Feature_tutorial.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Tutorial\n",
    "\n",
    "Besides datatracks, another aspect of chromatin/genome/structural biology which is routinely needed to be investigated are features. This can be in an extremely broad sense but to a first approximation a genome feature is just a named region of the genome which doesn't change. This is distinct from datatracks which define some function over genomic regions (see the Datatrack tutorial). \n",
    "\n",
    "Ofcourse, we could just encode features as numpy arrays on some chromosome:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('4', array([10, 20]))\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "myregion = np.array([10,20])\n",
    "mychrom = '4'\n",
    "\n",
    "myfeature = (mychrom, myregion)\n",
    "print(myfeature)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, this isn't the most flexible and when we scale to large numbers of features it may not even be so descriptive or useful. Firstly, we often need a way to link features together. For example:\n",
    "- A given gene may be associated with a given promoter. \n",
    "- A topologically associated domain (TAD) may contain multiple promoters within it\n",
    "\n",
    "Secondly, features may have condition-specific attributes which we might want to assign to them (while still having access to the same condition-independent feature). For example:\n",
    "\n",
    "- A promoter might have active marks in naive mESCs but bivalent marks in primed mESCs\n",
    "- A TAD may be within the A compartment in naive mESCs but move to the B compartment in primed mESCs\n",
    "\n",
    "Finally, we may want to keep track of datatracks within specific classes of features. For example:\n",
    "\n",
    "- We may want to know average rates of DNA methylation at active vs. bivalent promoters in naive mESCs\n",
    "\n",
    "In order to do all of these things, we'll make use of the Datatrack classes and the Feature classes. A bit of work will be needed to create our feature classes but once done it will provide some flexible analysis tools."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from GenTools import Feature as F\n",
    "from GenTools import Datatrack as Dt\n",
    "from GenTools.utils import dtrack_utils as dtu\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Feature Class\n",
    "The main class which i'll use to encode general features is the Feature class. At it's base, all a feature needs is a chromosome, some regions and a unique ID which allows us to keep track of that feature:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "This is a feature of with ID:\tUID1\n",
      "and with regions:\t[[10 20]]\n",
      "onchromosome:\t1\n"
     ]
    }
   ],
   "source": [
    "myfeat1 = F.Feature('UID1', '1', np.array([[10,20]]).astype('int32'))\n",
    "\n",
    "print(\"This is a feature of with ID:\\t{}\\nand with regions:\\t{}\\nonchromosome:\\t{}\".format(myfeat1.id, \n",
    "                                                                                         myfeat1.attrs['regions'],\n",
    "                                                                                         myfeat1.attrs['chromosome']\n",
    "                                                                                        ))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Attributes\n",
    "One optional input when we construct our feature is attrs (attributes). Essentially this is extra info beyond the feature ID and the feature type which we might want to associate with the feature. Let's add a 'chromosome' attribute to our feature myfeat1:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'regions': array([[10, 20]], dtype=int32), 'chromosome': '1', 'type': 'gene'}\n"
     ]
    }
   ],
   "source": [
    "myfeat1.attrs['type'] = 'gene'\n",
    "\n",
    "print(myfeat1.attrs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Feature attributes can really be anything:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'regions': array([[10, 20]], dtype=int32), 'chromosome': '1', 'type': 'gene', 'splicing sites': array([[15, 16],\n",
      "       [18, 19]], dtype=int32)}\n"
     ]
    }
   ],
   "source": [
    "myfeat1.attrs['splicing sites'] = np.array([[15,16],[18,19]]).astype('int32')\n",
    "print(myfeat1.attrs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Children\n",
    "Features can also have children:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Lets make a second feature of type 2\n",
    "myfeat2 = F.Feature('UID2',\n",
    "                    '1',\n",
    "                    np.array([[11, 13]]).astype('int32')\n",
    "                   )\n",
    "\n",
    "\n",
    "#Now let's add it to our first feature as a child\n",
    "myfeat1.add_child(myfeat2, #The child feature we want to add\n",
    "                  'type2' #The type of child feature we're adding\n",
    "                 )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The children of myfeat1 are stored in a dictionary in myfeat1.children. In particular, since we have specified our child type as 'type2', myfeat2 will appear in myfeat1.children['type2']:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Children of UID1:\n",
      "{'type2': [<GenTools.Feature.Feature object at 0x7fe3f5722190>]}\n",
      "Attributes of UID2\n",
      "{'regions': array([[11, 13]], dtype=int32), 'chromosome': '1'}\n"
     ]
    }
   ],
   "source": [
    "print(\"Children of {}:\".format(myfeat1.id))\n",
    "print(myfeat1.children)\n",
    "\n",
    "#We can also access the attributes of children\n",
    "print(\"Attributes of {}\".format(myfeat1.children['type2'][0].id))\n",
    "print(myfeat1.children['type2'][0].attrs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that at the moment there is nothing stopping the same feature from being added twice:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'type2': [<GenTools.Feature.Feature object at 0x7fe3f5722190>, <GenTools.Feature.Feature object at 0x7fe3f5722190>]}\n"
     ]
    }
   ],
   "source": [
    "#Now let's add it to our first feature as a child\n",
    "myfeat1.add_child(myfeat2, #The child feature we want to add\n",
    "                  'type2' #The type of child feature we're adding\n",
    "                 )\n",
    "\n",
    "print(myfeat1.children)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, since we have added the same feature (myfeat2) twice, note that the location of each item in the list myfeat1.children['type2'] is the same i.e. they're the same object. \n",
    "\n",
    "Let's see what happens if we add an attribute to our myfeat2:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before editing the region the attributes of UID2 are:\n",
      "{'regions': array([[11, 13]], dtype=int32), 'chromosome': '1'}\n",
      "\n",
      "\n",
      "###################################################\n",
      "After editing the region of the attributes of UID2, accessing the attrs via UID2 we get:\n",
      "{'regions': array([[11, 17]], dtype=int32), 'chromosome': '1'}\n",
      "\n",
      "\n",
      "###################################################\n",
      "After editing the region of the attributes of UID2, accessing the attrs via UID1 we get:\n",
      "Child type2 number 0:\n",
      "{'regions': array([[11, 17]], dtype=int32), 'chromosome': '1'}\n",
      "Child type2 number 1:\n",
      "{'regions': array([[11, 17]], dtype=int32), 'chromosome': '1'}\n"
     ]
    }
   ],
   "source": [
    "print(\"Before editing the region the attributes of {} are:\".format(myfeat2.id))\n",
    "print(myfeat2.attrs)\n",
    "print(\"\\n\")\n",
    "print(\"###################################################\")\n",
    "#Changing the region of myfeat2 to be longer\n",
    "myfeat2.attrs['regions'][0,1] = 17\n",
    "\n",
    "print(\"After editing the region of the attributes of {}, accessing the attrs via {} we get:\".format(myfeat2.id, myfeat2.id))\n",
    "print(myfeat2.attrs)\n",
    "print(\"\\n\")\n",
    "print(\"###################################################\")\n",
    "\n",
    "print(\"After editing the region of the attributes of {}, accessing the attrs via {} we get:\".format(myfeat2.id, myfeat1.id))\n",
    "print(\"Child type2 number 0:\")\n",
    "print(myfeat1.children['type2'][0].attrs)\n",
    "print(\"Child type2 number 1:\")\n",
    "print(myfeat1.children['type2'][1].attrs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the two added children (both myfeat2) are actually the same object, when we edit that object, the same edits are applied to the children within myfeat1. \n",
    "\n",
    "Finally, the input 'type2' into the add_child method is not entirely necessary since it can be inferred from myfeat2:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'type2': [<GenTools.Feature.Feature object at 0x7fe3f5719550>]}\n"
     ]
    }
   ],
   "source": [
    "myfeat1 = F.Feature('UID1', '1', np.array([[20,30]]).astype('int32'))\n",
    "myfeat2 = F.Feature('UID2', '1', np.array([[15,25]]).astype('int32'))\n",
    "myfeat1.add_child(myfeat2, 'type2')\n",
    "\n",
    "print(myfeat1.children)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, we may want to group children of the same type when we add children to our features. Therefore we can specify the child type:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'gene_class_a': [<GenTools.Feature.Feature object at 0x7fe3f572d5d0>], 'gene_class_b': [<GenTools.Feature.Feature object at 0x7fe3f5726bd0>]}\n"
     ]
    }
   ],
   "source": [
    "myfeat1 = F.Feature('UID1', '1', np.array([[20,30]]).astype('int32'))\n",
    "myfeat2 = F.Feature('UID2', '1', np.array([[15,25]]).astype('int32'), attrs = {'type': 'gene'})\n",
    "myfeat3 = F.Feature('UID3', '1', np.array([[26,32]]).astype('int32'), attrs = {'type': 'gene'}) #Third feature of the same type as myfeat2 but with a different UID\n",
    "myfeat1.add_child(myfeat2, 'gene_class_a')\n",
    "myfeat1.add_child(myfeat3, 'gene_class_b')\n",
    "\n",
    "print(myfeat1.children)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature Parents\n",
    "Features can also have parents. These work just the same as children:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Child information for UID1:\n",
      "UID2\n",
      "{'regions': array([[15, 25]], dtype=int32), 'chromosome': '1', 'type': 'gene'}\n",
      "\n",
      "\n",
      "Parent information for UID2:\n",
      "UID1\n",
      "{'regions': array([[20, 30]], dtype=int32), 'chromosome': '1', 'type': 'TAD'}\n"
     ]
    }
   ],
   "source": [
    "myfeat1 = F.Feature('UID1', '1', np.array([[20,30]]).astype('int32'), attrs = {'type':'TAD'})\n",
    "myfeat2 = F.Feature('UID2', '1', np.array([[15,25]]).astype('int32'), attrs = {'type': 'gene'})\n",
    "myfeat1.add_child(myfeat2, myfeat2.attrs['type'])\n",
    "myfeat2.add_parent(myfeat1, myfeat1.attrs['type'])\n",
    "\n",
    "print(\"Child information for {}:\".format(myfeat1.id))\n",
    "print(myfeat1.children['gene'][0].id)\n",
    "print(myfeat1.children['gene'][0].attrs)\n",
    "print(\"\\n\")\n",
    "print(\"Parent information for {}:\".format(myfeat2.id))\n",
    "print(myfeat2.parents['TAD'][0].id)\n",
    "print(myfeat2.parents['TAD'][0].attrs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Single Condition Features\n",
    "Sometimes we might want to associate a feature with a single condition and for this there is the Feature_single_condition class. The only difference between this and the standard Feature class is that the initialisation of one of these objects requires that a condition be specified. For example we might specify a TAD which seems to be unique to the naive mESC timepoint:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TAD_UID1\n",
      "naive\n",
      "2\n",
      "[[10 20]]\n"
     ]
    }
   ],
   "source": [
    "mESC_TAD = F.Feature_single_condition('TAD_UID1',  #Feature UID\n",
    "                                      'naive',     #Condition\n",
    "                                      '2',         #Chromosome\n",
    "                                      np.array([[10,20]]).astype('int32') #Region\n",
    "                                     )\n",
    "print(mESC_TAD.id)\n",
    "print(mESC_TAD.attrs['condition'])\n",
    "print(mESC_TAD.attrs['chromosome'])\n",
    "print(mESC_TAD.attrs['regions'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Single Cell Features\n",
    "As a subclass of single-condition features sometimes we might want to associate a feature with a single cell and for this there is the Feature_single_cell class. The only difference between this and the Feature_single_condition class is that the initialisation of one of these objects requires that a cell identifier be specified. For example we might specify a loop which seems to be unique to a single cell in the naive mESC timepoint:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loop_UID1\n",
      "naive\n",
      "cell_1\n",
      "2\n",
      "[[10 20]\n",
      " [50 60]]\n"
     ]
    }
   ],
   "source": [
    "mESC_loop = F.Feature_single_cell('loop_UID1',  #Feature UID\n",
    "                                  'naive',      #Condition\n",
    "                                  'cell_1',     #Cell ID\n",
    "                                  '2',          #Chromosome\n",
    "                                  np.array([[10,20],[50,60]]).astype('int32') #Region\n",
    "                                )\n",
    "print(mESC_loop.id)\n",
    "print(mESC_loop.attrs['condition'])\n",
    "print(mESC_loop.attrs['cell'])\n",
    "print(mESC_loop.attrs['chromosome'])\n",
    "print(mESC_loop.attrs['regions'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that in this case 'regions' arent confined to a single region - it just refers to the regions of interest in this feature i.e. the contact points. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A conrete example: TADs and Promoters\n",
    "Let's look at an example using TAD boundaries defined in Bonev et al (2017) assigning:\n",
    "\n",
    "- Compartmental score to specific TSS sites\n",
    "- Transcriptional start sites to specific TADs\n",
    "\n",
    "For this we'll use the GRCm38_p2 genome build for mouse and download the TSS sites for each gene (this is publically available data but also included in the tutorial_data). Since the main task will be converting given data into a Feature form to be used by this module, let's take a look at the data which we'll use:\n",
    "\n",
    "## Genes, Promoters and TSS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gene Start (bp)</th>\n",
       "      <th>Strand</th>\n",
       "      <th>Chromosome Name</th>\n",
       "      <th>Associated Gene Name</th>\n",
       "      <th>Transcript Start (bp)</th>\n",
       "      <th>Gene End (bp)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Ensembl Gene ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000095309</th>\n",
       "      <td>21272179</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>Vmn1r125</td>\n",
       "      <td>21272179</td>\n",
       "      <td>21273102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000000126</th>\n",
       "      <td>59306928</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>Wnt9a</td>\n",
       "      <td>59306930</td>\n",
       "      <td>59333552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000000126</th>\n",
       "      <td>59306928</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>Wnt9a</td>\n",
       "      <td>59306928</td>\n",
       "      <td>59333552</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000086196</th>\n",
       "      <td>60338188</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>Gm13571</td>\n",
       "      <td>60355352</td>\n",
       "      <td>60371234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000086196</th>\n",
       "      <td>60338188</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>Gm13571</td>\n",
       "      <td>60338188</td>\n",
       "      <td>60371234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000077593</th>\n",
       "      <td>77682614</td>\n",
       "      <td>1</td>\n",
       "      <td>18</td>\n",
       "      <td>Gm26014</td>\n",
       "      <td>77682614</td>\n",
       "      <td>77682745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000065122</th>\n",
       "      <td>84649496</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>Gm26126</td>\n",
       "      <td>84649496</td>\n",
       "      <td>84649659</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000093072</th>\n",
       "      <td>114580495</td>\n",
       "      <td>1</td>\n",
       "      <td>13</td>\n",
       "      <td>Gm25950</td>\n",
       "      <td>114580495</td>\n",
       "      <td>114580594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000088303</th>\n",
       "      <td>118164078</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>Gm23382</td>\n",
       "      <td>118164078</td>\n",
       "      <td>118164181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000084441</th>\n",
       "      <td>110732036</td>\n",
       "      <td>1</td>\n",
       "      <td>14</td>\n",
       "      <td>Gm26255</td>\n",
       "      <td>110732036</td>\n",
       "      <td>110732138</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>94084 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Gene Start (bp)  Strand Chromosome Name  \\\n",
       "Ensembl Gene ID                                               \n",
       "ENSMUSG00000095309         21272179       1               7   \n",
       "ENSMUSG00000000126         59306928       1              11   \n",
       "ENSMUSG00000000126         59306928       1              11   \n",
       "ENSMUSG00000086196         60338188       1               2   \n",
       "ENSMUSG00000086196         60338188       1               2   \n",
       "...                             ...     ...             ...   \n",
       "ENSMUSG00000077593         77682614       1              18   \n",
       "ENSMUSG00000065122         84649496       1               9   \n",
       "ENSMUSG00000093072        114580495       1              13   \n",
       "ENSMUSG00000088303        118164078       1               4   \n",
       "ENSMUSG00000084441        110732036       1              14   \n",
       "\n",
       "                   Associated Gene Name  Transcript Start (bp)  Gene End (bp)  \n",
       "Ensembl Gene ID                                                                \n",
       "ENSMUSG00000095309             Vmn1r125               21272179       21273102  \n",
       "ENSMUSG00000000126                Wnt9a               59306930       59333552  \n",
       "ENSMUSG00000000126                Wnt9a               59306928       59333552  \n",
       "ENSMUSG00000086196              Gm13571               60355352       60371234  \n",
       "ENSMUSG00000086196              Gm13571               60338188       60371234  \n",
       "...                                 ...                    ...            ...  \n",
       "ENSMUSG00000077593              Gm26014               77682614       77682745  \n",
       "ENSMUSG00000065122              Gm26126               84649496       84649659  \n",
       "ENSMUSG00000093072              Gm25950              114580495      114580594  \n",
       "ENSMUSG00000088303              Gm23382              118164078      118164181  \n",
       "ENSMUSG00000084441              Gm26255              110732036      110732138  \n",
       "\n",
       "[94084 rows x 6 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# TRANSCRIPTONAL START SITES #################################################################################\n",
    "TSS_file = \"tutorial_data/ensembl_TSS_sites_GRCm38_p2.txt\"\n",
    "TSS = pd.read_csv(TSS_file, sep = '\\t', index_col = 0)\n",
    "TSS"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see there are A LOT of rows (94084!). Let's try and pull out the Nanog gene (ensembl ID: ENSMUSG00000012396):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gene Start (bp)</th>\n",
       "      <th>Strand</th>\n",
       "      <th>Chromosome Name</th>\n",
       "      <th>Associated Gene Name</th>\n",
       "      <th>Transcript Start (bp)</th>\n",
       "      <th>Gene End (bp)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Ensembl Gene ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000012396</th>\n",
       "      <td>122707489</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>Nanog</td>\n",
       "      <td>122707489</td>\n",
       "      <td>122714633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000012396</th>\n",
       "      <td>122707489</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>Nanog</td>\n",
       "      <td>122707565</td>\n",
       "      <td>122714633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ENSMUSG00000012396</th>\n",
       "      <td>122707489</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>Nanog</td>\n",
       "      <td>122707568</td>\n",
       "      <td>122714633</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    Gene Start (bp)  Strand Chromosome Name  \\\n",
       "Ensembl Gene ID                                               \n",
       "ENSMUSG00000012396        122707489       1               6   \n",
       "ENSMUSG00000012396        122707489       1               6   \n",
       "ENSMUSG00000012396        122707489       1               6   \n",
       "\n",
       "                   Associated Gene Name  Transcript Start (bp)  Gene End (bp)  \n",
       "Ensembl Gene ID                                                                \n",
       "ENSMUSG00000012396                Nanog              122707489      122714633  \n",
       "ENSMUSG00000012396                Nanog              122707565      122714633  \n",
       "ENSMUSG00000012396                Nanog              122707568      122714633  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "TSS.loc['ENSMUSG00000012396']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that there are potentially multiple TSS sites per gene! It's kind of hard to visualise them like this so let's look at the differences in basepairs between the start sites:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([76,  3])"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.diff(TSS.loc['ENSMUSG00000012396','Transcript Start (bp)'].values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the case of Nanog, the TSS sites are so close that it seems entirely feasible these TSS sites share the same promoter region. However, this may not be the case for all genes. Let's take a look at the distribution of TSS differences across genes with mutliple TSS:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "diffs = np.empty((1,0))\n",
    "tot = 0\n",
    "for ID in list(set(TSS.index.values)):\n",
    "    if tot == 999:\n",
    "        break\n",
    "    try:\n",
    "        vals = TSS.loc[ID,'Transcript Start (bp)'].values\n",
    "        diffs = np.append(diffs, np.diff(vals))\n",
    "        tot += 1\n",
    "    except:\n",
    "        #Only one TSS site\n",
    "        continue"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Genomic distance between TSS sites\\nfor 1000 genes with multiple TSS sites')"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize = (10,10))\n",
    "hist = plt.hist(np.log10(diffs[diffs>0]))\n",
    "plt.xlabel(\"log10 of genomic distance\", size = 20)\n",
    "plt.ylabel(\"Frequency\", size = 20)\n",
    "plt.tick_params(labelsize = 15)\n",
    "plt.title(\"Genomic distance between TSS sites\\nfor {} genes with multiple TSS sites\".format(tot+1), size = 20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It looks like some TSS sites can be $\\textbf{very}$ far away from each other! To simplify things let's just take the top hit for each ensembl ID (Given the way that gene information is downloaded from biomart this will correspond to the transcript {gene_name}_001:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Gene(F.Feature):\n",
    "    def __init__(self,\n",
    "                 fid,\n",
    "                 chromosome,\n",
    "                 genebody,\n",
    "                 strand,\n",
    "                 promoter,\n",
    "                 attrs = None,\n",
    "                 parents = None,\n",
    "                 children = None):\n",
    "        super(Gene,self).__init__(fid,\n",
    "                                  chromosome,\n",
    "                                  genebody,\n",
    "                                  attrs = attrs,\n",
    "                                  parents= parents,\n",
    "                                  children = children)\n",
    "        self.attrs['strand'] = strand\n",
    "        self.attrs['promoter'] = promoter"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we've defined our class with some features we might be interested in, let's create a list of gene features which we'll use to link with our TADs:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_gene_and_prom_obs(ID):\n",
    "    '''\n",
    "    Function to construct a Gene object from our TSS table given some ensembl ID\n",
    "    '''\n",
    "    gene = TSS.loc[ID]\n",
    "    \n",
    "    #If there is more than one transcript we want to pick the top one!\n",
    "    if len(gene.values.shape)>1:\n",
    "        start = gene['Gene Start (bp)'].values[0]\n",
    "        end = gene['Gene End (bp)'].values[0]\n",
    "        strand = gene['Strand'].values[0]\n",
    "        mychrom = gene['Chromosome Name'].values[0]\n",
    "        name = gene['Associated Gene Name'].values[0]\n",
    "    else:\n",
    "        start = gene['Gene Start (bp)']\n",
    "        end = gene['Gene End (bp)']\n",
    "        strand = gene['Strand']\n",
    "        mychrom = gene['Chromosome Name']\n",
    "        name = gene['Associated Gene Name']\n",
    "\n",
    "    \n",
    "    #Create our genebody region\n",
    "    genebody = [start,end]\n",
    "    #Making the assumption that the promoter region will be somewhere within 2kb\n",
    "    #of the TSS (upstream or downstream depends on the strand information)\n",
    "    if strand == 1:\n",
    "        myprom = [start - int(2e3), start]\n",
    "    else:\n",
    "        myprom = [end, end + int(2e3)]\n",
    "\n",
    "    #Instantiate a new gene object\n",
    "    mygene = Gene(ID,\n",
    "                  mychrom,\n",
    "                  genebody,\n",
    "                  strand,\n",
    "                  promoter = myprom,\n",
    "                  attrs = {'name': name})\n",
    "                  \n",
    "    return mychrom, mygene\n",
    "\n",
    "'''\n",
    "Using some multiprocessing here to whizz through all the ensembl IDs\n",
    "All this is really doing is working in multiple threads to create a\n",
    "gene object for each gene and store the result in a dictionary \n",
    "with chromosomes as keys\n",
    "'''\n",
    "chroms = [str(i+1) for i in np.arange(19)]+['X'] \n",
    "genes = {chrom: [] for chrom in chroms}\n",
    "myIDs = list(set(TSS.index.values))\n",
    "\n",
    "from multiprocessing import Pool\n",
    "p = Pool()\n",
    "temp_outputs = p.imap(make_gene_and_prom_obs, (ID for ID in myIDs))\n",
    "for temp_output in temp_outputs: \n",
    "    genes[temp_output[0]].append(temp_output[1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now access gene information by accessing the dictionary of gene objects:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ENSMUSG00000067750\n",
      "\n",
      "\n",
      "[21349641, 21352200]\n",
      "\n",
      "\n",
      "[21347641, 21349641]\n"
     ]
    }
   ],
   "source": [
    "print(genes['1'][10].id)\n",
    "print(\"\\n\")\n",
    "print(genes['1'][10].attrs['regions'])\n",
    "print(\"\\n\")\n",
    "print(genes['1'][10].attrs['promoter'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Naive mESC TADs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>chrom</th>\n",
       "      <th>start</th>\n",
       "      <th>end</th>\n",
       "      <th>Compartment</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>chr1</td>\n",
       "      <td>0</td>\n",
       "      <td>4328000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>chr1</td>\n",
       "      <td>4461000</td>\n",
       "      <td>4761000</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>chr1</td>\n",
       "      <td>4782000</td>\n",
       "      <td>5072000</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>chr1</td>\n",
       "      <td>5089000</td>\n",
       "      <td>6980000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>chr1</td>\n",
       "      <td>7003000</td>\n",
       "      <td>9437000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2707</th>\n",
       "      <td>chrY</td>\n",
       "      <td>5300000</td>\n",
       "      <td>5426000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2708</th>\n",
       "      <td>chrY</td>\n",
       "      <td>5432000</td>\n",
       "      <td>6658000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2709</th>\n",
       "      <td>chrY</td>\n",
       "      <td>6665000</td>\n",
       "      <td>6751000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2710</th>\n",
       "      <td>chrY</td>\n",
       "      <td>6792000</td>\n",
       "      <td>9788000</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2711</th>\n",
       "      <td>chrY</td>\n",
       "      <td>9905000</td>\n",
       "      <td>91744698</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2712 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     chrom    start       end Compartment\n",
       "0     chr1        0   4328000           B\n",
       "1     chr1  4461000   4761000           A\n",
       "2     chr1  4782000   5072000           A\n",
       "3     chr1  5089000   6980000           B\n",
       "4     chr1  7003000   9437000           B\n",
       "...    ...      ...       ...         ...\n",
       "2707  chrY  5300000   5426000           B\n",
       "2708  chrY  5432000   6658000           B\n",
       "2709  chrY  6665000   6751000           B\n",
       "2710  chrY  6792000   9788000           B\n",
       "2711  chrY  9905000  91744698           B\n",
       "\n",
       "[2712 rows x 4 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "TADs = pd.read_csv(\"tutorial_data/TAD_boundaries_mESC.csv\",sep = \",\")\n",
    "TADs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's go ahead and make a TAD class to capture this information:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TAD(F.Feature_single_condition):\n",
    "    '''\n",
    "    Base class to represent a TAD\n",
    "    '''\n",
    "    def __init__(self,\n",
    "                 fid,\n",
    "                 condition,\n",
    "                 chromosome,\n",
    "                 regions,\n",
    "                 attrs = None,\n",
    "                 parents = None,\n",
    "                 children = None):\n",
    "                         \n",
    "        super(TAD,self).__init__(fid,\n",
    "                                 condition,\n",
    "                                 chromosome,\n",
    "                                 regions,\n",
    "                                 attrs = attrs,\n",
    "                                 parents= parents,\n",
    "                                 children = children)                    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that we've used:\n",
    "\n",
    "- A subclass of Feature_single_condition since TADs are condtion specific\n",
    "- To specify each TAD all we need is a region and a chromosome so we use those as attributes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "TAD_dict = {chrom: [] for chrom in chroms}\n",
    "for idx in TADs.index.values:\n",
    "    myTAD = TADs.loc[idx]\n",
    "    mychrom = myTAD['chrom'][3:]\n",
    "    if mychrom == 'Y':\n",
    "        continue\n",
    "    myregion = [myTAD['start'], myTAD['end']]\n",
    "    \n",
    "    myID = \"TAD_{}\".format(idx)\n",
    "    \n",
    "    TAD_dict[mychrom].append(TAD(myID,\n",
    "                                 'naive',\n",
    "                                 mychrom,\n",
    "                                 myregion\n",
    "                                )\n",
    "                            )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now access TAD information as TAD objects:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TAD_10\n",
      "\n",
      "\n",
      "[15026000, 17097000]\n",
      "\n",
      "\n",
      "1\n"
     ]
    }
   ],
   "source": [
    "print(TAD_dict['1'][10].id)\n",
    "print(\"\\n\")\n",
    "print(TAD_dict['1'][10].attrs['regions'])\n",
    "print(\"\\n\")\n",
    "print(TAD_dict['1'][10].attrs['chromosome'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Linking TADs and Genes\n",
    "\n",
    "We only really gain anything from this if we can efficiently link together different sets of features. 'Linking' could be applied very broadly and could be used to add children features of type1 to parent features type2 where feature A of type 1 is a child of feature B of type 2 if:\n",
    "\n",
    "- their genomic regions overlap\n",
    "- they have similar binding of a protein\n",
    "- Feature A is regulated by Feature 2\n",
    "\n",
    "But how do we go about linking features? The Feature module contains a function link_features described here:\n",
    "\n",
    "    '''\n",
    "    Given some list of regions associated with each parent and child, this returns an (N,2) shape\n",
    "    array detailing the links between parents and children. \n",
    "    Arguments:\n",
    "    parent_regions: A list of (M,2) shape arrays (one per parent) which need not be the same length\n",
    "                    but which detail the regions associated with each parent.\n",
    "    child_regions: A list of (M,2) shape arrays (one per child) which need not be the same length\n",
    "                   but which detail the regions associated with each child.\n",
    "    \n",
    "    Returns:\n",
    "    Links: (N,2) shape array (essentially a COO format sparse matrix) where each row details \n",
    "           a link between a parent and a child. \n",
    "    '''\n",
    "    \n",
    "In this tutorial we'll just look at associating parent features with child features using genomic overlap. Given this, each gene and TAD will have to be associated with a set of regions. We provide the function with a list of these regions for the parent sets and child sets and the function tells us which members of the parent and child sets overlap.\n",
    "\n",
    "### TAD overlap with gene bodies\n",
    "\n",
    "#### Chromosome 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "gene_regions = [np.array(item.attrs['regions']).reshape(1,2).astype('int32') for item in genes['1']]\n",
    "TAD_regions = [np.array(item.attrs['regions']).reshape(1,2).astype('int32') for item in TAD_dict['1']]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We've therefore associated each gene and TAD with a region encoded as a shape (1,2) integer array. If we now call link_features then this will return an array where each row details a TAD (parent) - gene (child) pair: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[   0  210]\n",
      " [   0  529]\n",
      " [   0  783]\n",
      " ...\n",
      " [ 195 1650]\n",
      " [ 195 1713]\n",
      " [ 195 1918]]\n"
     ]
    }
   ],
   "source": [
    "links = F.link_features(TAD_regions,\n",
    "                        gene_regions)\n",
    "print(links)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that this is fairly quick as long as each parent and child is associated with a single region. However, it is possible to associate a Feature with multiple regions - link features will still work in this case, it will just be slower! Anyway, let's go ahead and add the gene children to the parent TADs:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "for link in links:\n",
    "    TAD_dict['1'][link[0]].add_child(genes['1'][link[1]], 'genebody')\n",
    "    genes['1'][link[1]].add_parent(TAD_dict['1'][link[0]], 'genebody_TAD')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can identify for each TAD, the genes which are overlapped with that TAD:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['ENSMUSG00000042686', 'ENSMUSG00000073737', 'ENSMUSG00000089982', 'ENSMUSG00000091020', 'ENSMUSG00000067795', 'ENSMUSG00000088943', 'ENSMUSG00000025920', 'ENSMUSG00000077377', 'ENSMUSG00000032719', 'ENSMUSG00000025921', 'ENSMUSG00000090498', 'ENSMUSG00000097744', 'ENSMUSG00000079658', 'ENSMUSG00000088829', 'ENSMUSG00000025779', 'ENSMUSG00000025940', 'ENSMUSG00000092083', 'ENSMUSG00000025939', 'ENSMUSG00000025925', 'ENSMUSG00000043716']\n"
     ]
    }
   ],
   "source": [
    "print([item.id for item in TAD_dict['1'][10].children['genebody']])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note that above, when we have called add_child and add_parent we have specified that these parent-child links are associated wiht the genebody. We can do a similar analysis for gene promoter regions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "gene_regions = [np.array(item.attrs['promoter']).reshape(1,2).astype('int32') for item in genes['1']]\n",
    "TAD_regions = [np.array(item.attrs['regions']).reshape(1,2).astype('int32') for item in TAD_dict['1']]\n",
    "\n",
    "links = F.link_features(TAD_regions,\n",
    "                        gene_regions)\n",
    "\n",
    "for link in links:\n",
    "    TAD_dict['1'][link[0]].add_child(genes['1'][link[1]], 'promoter')\n",
    "    genes['1'][link[1]].add_parent(TAD_dict['1'][link[0]], 'promoter_TAD')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is very useful! Let's now take a look at the distribution of the number of promoters in each TAD, scaled by the length of that TAD:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Promoter density per TAD')"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "promdensity_chr1 = [len(item.children['promoter'])/(item.attrs['regions'][1] - item.attrs['regions'][0]) if 'promoter' in item.children else 0 for item in TAD_dict['1']]\n",
    "\n",
    "fig = plt.figure(figsize = (10,10))\n",
    "hist = plt.hist(promdensity_chr1)\n",
    "plt.xlabel(\"Promoter Density (bp^-1)\", size = 20)\n",
    "plt.ylabel(\"Frequency\", size = 20)\n",
    "plt.tick_params(labelsize = 12)\n",
    "plt.title(\"Promoter density per TAD\".format(tot+1), size = 20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Combining with the Datatrack module\n",
    "This is all very nice but really what we want is simple ways to manipulate features alongside Datatracks! Let's do that here. Suppose we wanted to know whether Nanog binding to promoters shows significant correlation for genes within the same TAD. Let's go ahead and investigate this now.\n",
    "\n",
    "First things first, we need to load our Nanog datatrack:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "nanog_file = \"tutorial_data/GSEXXXXXHendrich20161026_Nanog_ESC_peaks.npz\"\n",
    "nanog = Dt.DataTrack_rvp(\"Nanog\").from_npz(nanog_file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now link TADs and promoters for all genes:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Clear out our previous children and parents\n",
    "for chrom in chroms:\n",
    "    for TAD in TAD_dict[chrom]:\n",
    "        if 'promoter' in TAD.children:\n",
    "            del(TAD.children['promoter'])\n",
    "    for gene in genes[chrom]:\n",
    "        if 'promoter_TAD' in gene.parents:\n",
    "            del(gene.parents['promoter_TAD'])\n",
    "    \n",
    "for chrom in chroms:\n",
    "    prom_regions = [np.array(item.attrs['promoter']).reshape(1,2).astype('int32') for item in genes[chrom]]\n",
    "    TAD_regions = [np.array(item.attrs['regions']).reshape(1,2).astype('int32') for item in TAD_dict[chrom]]\n",
    "\n",
    "    links = F.link_features(TAD_regions,\n",
    "                            prom_regions)\n",
    "\n",
    "    for link in links:\n",
    "        TAD_dict[chrom][link[0]].add_child(genes[chrom][link[1]], 'promoter')\n",
    "        genes[chrom][link[1]].add_parent(TAD_dict[chrom][link[0]], 'promoter_TAD')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's assign some Nanog attributes to our promoters:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "for chrom in chroms:\n",
    "    gene_regions = [np.array(item.attrs['promoter']).reshape(1,2).astype('int32') for item in genes[chrom]]\n",
    "    max_nanog = nanog.stats(chrom, gene_regions, stats_type = 'max')\n",
    "    nanog_coverage = nanog.stats(chrom,gene_regions, stats_type = 'coverage')\n",
    "    \n",
    "    for idx in np.arange(len(max_nanog)):\n",
    "        genes[chrom][idx].attrs['promoter_maximum_nanog_signal'] = max_nanog[idx]\n",
    "        genes[chrom][idx].attrs['promoter_nanog_coverage'] = nanog_coverage[idx]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now let's look at the distributions of within-promoter maximum Nanog signal (WP-MNS) in each TAD. In particular lets look at the mean WP-MNS versus the standard deivation in WP-MNS. If binding of Nanog to one promoter in a TAD raised the probability of Nanog binding to another promoter in the TAD we would expect the standard deviation in Nanog binding to stay relatively constant when compared with the mean. Let's see if this is what we see:\n",
    "\n",
    "### Maximum nanog promoter signal for genes in same TAD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fe3e028e510>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "means = []\n",
    "std = []\n",
    "nums = []\n",
    "for chrom in chroms:\n",
    "    for TAD in TAD_dict[chrom]:\n",
    "        if 'promoter' in TAD.children:\n",
    "            proms = TAD.children['promoter']\n",
    "            nums.append(len(proms))\n",
    "            means.append(np.mean([item.attrs['promoter_maximum_nanog_signal'] for item in proms]))\n",
    "            std.append(np.std([[item.attrs['promoter_maximum_nanog_signal']] for item in proms]))\n",
    "\n",
    "fig = plt.figure(figsize = (10,10))\n",
    "plt.scatter(means, std, c = nums, cmap = plt.cm.Blues, vmin = -10, vmax = 50, label = \"TAD\\nColor = Number of Promoters\")\n",
    "plt.xscale('symlog')\n",
    "plt.yscale('symlog')\n",
    "plt.ylim([0,5e3])\n",
    "plt.tick_params(labelsize = 15)\n",
    "plt.xlabel(\"Cross-promoter mean of\\nwithin-promoter maximum Nanog signal\", size = 20)\n",
    "plt.ylabel(\"Cross-promoter standard-deviation of\\nwithin-promoter maximum Nanog signal\", size = 20)\n",
    "plt.legend(prop = {'size':15})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It looks as if we have a noisy linear relationship between the mean and the variance of within-promoter maximum Nanog signal (WP-MN).\n",
    "\n",
    "It also seems like for a higher mean WP-MN is associated with lower numbers of promoters. This could just be down to the specificity of Nanog binding - if Nanog only binds 5% of promoters then for a TAD with lots of promoters, the within-TAD mean WP-MN will be dragged down my most promoters having no Nanog binding. Therefore we don't see this trend in the within-TAD standard deivation of WP-MN.\n",
    "\n",
    "It looks as if Nanog doesn't show huge amount of TAD specificity! Let's double check this by looking at a histogram of mean WP-MNS within each TAD:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Frequency')"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAm0AAAJbCAYAAAC2KKMOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3de7xl93z/8dfbTIIhN8mQIukgyM+tfuTHr6FC4tJI8kPxK6WqSkqrPFA1NCEoiRZtJS6JW/ipuvYiiQiRmwglcalLElGGEEF0iEnkNvn8/ljryM7OPnPO2XNm7/M95/V8PPZjz1nru9b+nDVrzrzP97vWd6WqkCRJ0tJ2s2kXIEmSpLkZ2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIasHraBWxru+22W61bt27aZUiSJM3pvPPOu6yq1o5at+xD27p16zj33HOnXYYkSdKcknxvtnUOj0qSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ1YPe0CJGmUdetPmthnbTjqoIl9liSNy542SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWrA6mkXIGnrrVt/0sQ+a8NRB03ssyRJN7CnTZIkqQFTD21JVidZn+SiJFcn+UGSvx9qkyQvS3Jxkl8lOSvJfadVsyRJ0qQtheHRdwMHAK8ELgD2AO4x1GY9cDjw4r7NC4FTk9yrqi6dYK2SJElTMdXQluR3gScBv1VV35ylzS3oQtuRVXVMv+xzwAbgucBhk6lWkiRpeqY9PPoM4LTZAltvX2BH4EMzC6rqCuAE4MBtW54kSdLSMO3Q9kDgW0mOSXJ5kiuT/EuS2w+02RvYDFw0tO35/TpJkqRlb9qhbXfg6cB96YZJ/xi4P/CvSdK32QXYVFWbh7bdCKxJsv2EapUkSZqaad+IkP71mKr6GUCSHwFnAvsDn+7b1SzbjlyX5FDgUIA999xzkUuWJEmavGn3tG0EvjYT2HpnA9dwwx2kG4Edkqwa2nZn4MqqunZ4p1V1XFXtU1X7rF27dlvULUmSNFHTDm3nz7I8wPX9ny8AVgF7DbXZu18nSZK07E07tJ0I3CfJbgPLHgJsB3y1//oc4HLgiTMNkqwBDgFOnlCdkiRJUzXta9qOA54HnJDktcAOwOuAU6vqbICquirJUcDhSTZyw+S6NwOOnk7ZkiRJkzXV0FZVlyfZH3gT8AG6a9n+HXjBUNOj6ELaS4FdgXOBR1TVjydYriRJ0tRMu6eNqvo28Og52hTwmv4lSZK04kz7mjZJkiTNw9R72iRp2tatP2lin7XhqIMm9lmSlhd72iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAc7TJmlBJjmnmSTpBva0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1YOqhLcnTk9SI17MH2iTJy5JcnORXSc5Kct9p1i1JkjRJq6ddwID9gV8NfP2dgT+vBw4HXgxcALwQODXJvarq0smVKEmSNB1LKbR9sao2DS9Mcgu60HZkVR3TL/scsAF4LnDYJIuUJEmahqkPj87DvsCOwIdmFlTVFcAJwIHTKkqSJGmSllJo+68k1yW5MMmfDizfG9gMXDTU/vx+nSRJ0rK3FIZHf0R3vdoXgFXAk4G3JVlTVX8P7AJsqqrNQ9ttBNYk2b6qrploxZIkSRM29dBWVacApwwsOjnJzYHDkvzjTLMRm2a2dUkOBQ4F2HPPPRexWkmSpOlYSsOjgz4C3AZYR9ejtkOSVUNtdgaurKprhzeuquOqap+q2mft2rXbvFhJkqRtbamGthlFN8XHKmCvoXV79+skSZKWvaUa2h4PXAZ8DzgHuBx44szKJGuAQ4CTp1KdJEnShE39mrYkH6W7CeE/6XrUfr9/Pa+qrgeuSnIUcHiSjdwwue7NgKOnU7UkSdJkTT20ARcCzwD2oLu54JvA06rq/w20OYoupL0U2BU4F3hEVf14wrVKkiRNxdRDW1W9DHjZHG0KeE3/kiRJWnGW6jVtkiRJGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAasnnYBkrSSrFt/0kQ+Z8NRB03kcyRNjj1tkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktSA1dMuQFqu1q0/adolSJKWEXvaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhqwpEJbkjsk2ZSkktx6YHmSvCzJxUl+leSsJPedZq2SJEmTtKRCG/B3wKYRy9cDhwOvAw7p25yaZPcJ1iZJkjQ1Sya0Jfkd4HeB1w8tvwVdaDuyqo6pqlOBJwIFPHfihUqSJE3BgkNbktWLXUSSVcDRwKuAy4ZW7wvsCHxoZkFVXQGcABy42LVIkiQtReP0tF2c5DVJ7rSIdTwbuAXw5hHr9gY2AxcNLT+/XydJkrTsjRPabg68FLgoyclJHpNk7GHWJLsCrwZeWFXXjmiyC7CpqjYPLd8IrEmy/bifLUmS1IpxwtZvAE8HPg88CvgXut63VybZY4z9vQb4j6r6+Bba1IhlmW1dkkOTnJvk3J/+9KdjlCRJkrS0LDi0VdXVVfXeqnowcE/gGLqhzcOB7yT5WJKDkmSLOwKS3BN4BvDKJDsn2RlY06/eKckt6XrUduivexu0M3DlqN65qjquqvapqn3Wrl270G9RkiRpydmqu0er6vyqej5we27ofTsY+BiwIclhSW63hV3cFdgO+BxdONvIDde1/YDu5oQLgFXAXkPb7t2vkyRJWvYWZcqPqroa+Ffgn4Ef0g1d7kF3N+iGJK+f5dqzs4GHDb1e1697NN28becAl9NN8wFAkjV087WdvBj1S5IkLXVbPX1Hkn2APwWeRDe0uQl4C/Au4H7AC/rXLRiaV62qLgPOGNrfuv6Pn6mqTf2yo4DDk2yk6117IV3gPHpr65ckSWrBWKEtya2Ap9CFtfvS9az9J/BW4H39PGoAX0rybuCTwO8z/mS4R9GFtJcCuwLnAo+oqh+PuT9JkqSmLDi0JXkb8GTg1sC1dEOib6mqc0a1r6rNSU4DHjqf/VfV8cDxQ8uK7i7T1yy0XkmSpOVgnJ62Q4ENwGuBd/ZDnHM5s28vSZKkMYwT2g4BPt73fs1LVZ1Nd9OBJEmSxrDg0FZVJ22LQiRJkjS7cR4Y/7AkxyX5jVnW375f/5CtL0+SJEkw3vDo84B7VtWPRq2sqkv6wLYLcNbWFCdJkqTOOJPr3p+5r087G3jAGPuWJEnSCOOEttsCl8zR5tK+nSRJkhbBOKHtF8Ad52hzR+CKOdpIkiRpnsYJbV8EHjvbg+CT7A48tm8nSZKkRTBOaDsG2BE4K8mjk6wGSLI6yUF0E+nugM8FlSRJWjTjzNP2iSRH0j0H9ATg+iSXAbvRhcAAR1bVxxe1UkmSpBVsnJ42quqvgYPpHgT/S7qbDn4JnAIc1K+XJEnSIhlnnjYA+p40e9MkSZImYKyeNkmSJE3W2D1tAEluDuwMrBq1vqrmms9NkiRJ8zBWaEvyZOAlwL3objwYpcbdvyRJkm5swaEqyR8C7wGuBz4PXAxct8h1SZIkacA4PWF/RfdUhN+pqq8vcj2SJEkaYZwbEe4KfMjAJkmSNDnjhLaNwK8WuxBJkiTNbpzQdhLw0CSz3YAgSZKkRTZOaFsP3Ap4c5I1i1yPJEmSRhjnRoT3092I8KfAU5JcCPx8RLuqqkdtTXGSJEnqjBPaHj7w5x2AfWZpV2PsW5IkSSOME9q2W/QqJEmStEULDm1VtXlbFCJNwrr1J027BEmSxuID4yVJkhowVmhL5zlJzk7ysyRXDay7b5I3Jbnr4pUpSZK0si04tCXZDjgFOAa4B3A1N77O7XvAocBTFqNASZIkjdfT9pd0d5D+DbAWOG5wZVVtBD4DON2HJEnSIhkntD0V+FxVvaK/KWHU1B7fAX5zqyqTJEnSr40T2u4MnDNHm/8Gdh1j35IkSRphnNB2FbDTHG32ZPRTEiRJkjSGcULbV4BHJNl+1MokOwKPBL6wNYVJkiTpBuOEtnfQXa/2niS3HlzRB7Z3AbcBjt368iRJkgTjPRHhn5I8EvhD4LHARoAknwfuDdwSOLaqTlzMQiVJklaysSbXrao/opuL7dvA7kCABwDfB/60qp6zaBVKkiRprAfGA1BV7wDe0Q+R3gb4RVX9YtEqkyRJ0q+NHdpmVNUmYNMi1CJJkqRZ+MB4SZKkBiy4py3Jt+bZtKrq7gvdvyRJkm5qnOHRNYx+dNVOwMwUID8Grhu3KEmSJN3YOFN+3HG2dUn2Bv4R2A44cCvqkiRJ0oBFvaatqi4AHgesAw5fzH1LkiStZIt+I0JVXQmcAjx1sfctSZK0Um2ru0evpZt0V5IkSYtg0UNbktvQDZH+YLH3LUmStFKNM+XHy7awrz3oAtsuwGFbUZckSZIGjDPlx9/MsX4TcFRVHTnGviVJkjTCOKHtEbMsvx7YCHyzqq4ZvyRJkiQNG2eetk9vi0IkSZI0O589KkmS1IBxbkS4/bgfVlWXjLutJEnSSjbONW0/YPSzR+dSY36eJEnSijdOiHo/sCfwYOCXwH8Cl9JNpnsfYAfgM8D3F6lGSZKkFW+c0PZK4HPA0cArqurnMyuS7Ay8Gngy8CdV9e1FqVKSJGmFG+dGhNcB51fV8wcDG0BV/byq/gK4oG8nSZKkRTBOaNsPOGuONmf17SRJkrQIxgltNwduN0eb3YFbjLFvSZIkjTBOaPsq8KQk9xm1Msl9gd8Hvrw1hUmSJOkG49yI8CrgJOALSd5LNxT6Y7ret/2AP+z3+6rFKlKSJGmlG+cxVqckeQrwNuCZwJ8MrA7wC+DZVfWpxSlRkiRJY012W1UfTPJx4HHA/YCd6MLal4B/rapfLl6JkiRJGvsJBX0we2//kiRJ0ja01Q+MT7JDkt9YjGIkSZI02lihLcmtkrwuyQ+AnwMXD6x7QJKP9XeRSpIkaREseHg0yQ7A2cC9ga8DlwN3H2jyDWB/uqcifGURapQkSVrxxulpO4wusD2zqu4DfGhwZVVdAZwJHDDXjpI8Ick5SX6W5KokFyY5LMn2A22S5GVJLk7yqyRn2YsnSZJWmnFC2+OBT1bVu/qva0SbDcAd57GvXYHT6aYOORB4F/DXwBsH2qwHDqd7lukhwCbg1CS7j1G7JElSk8a5e/SOwEfnaLOJbhqQLaqqY4cWnZ5kR+DPk/wF3SOz1gNHVtUxAEk+RxcKn0vX6ydJkrTsjdPTtglYO0ebOwGXjbFvgJ8BM8Oj+wI7MjAE2w+/nkDXMydJkrQijBPavggcnOTWo1b2w5YHAufMd4dJViVZk+TBwPOAt1ZVAXsDm4GLhjY5v18nSZK0IowT2t4E7AacmOSugyv6rz8I3LJvN19X9K/P0N3E8OJ++S7ApqraPNR+I7Bm8IYFSZKk5WycZ4+enORv6K4nuwC4GiDJpXTDpgH+uqrOXsBu9wXWAA8AXg4cA/zZzEeOaJ8trCPJocChAHvuuecCypCk5WHd+pMm9lkbjjpoYp8lrWRjTa5bVS8HHgV8nK6HDLqbBj4JPKqqjlzg/r5UVWdX1Rvphkefk+QudD1qOyRZNbTJzsCVVXXtLPs7rqr2qap91q6d6/I7SZKkpW9rnj36KeBTi1jLjC/173ei68lbBewFXDjQZu9+nSRJ0oqw4J62JJ9McsQ2qGXGg/r379LdzHA58MSBz19DN1/byduwBkmSpCVlnJ62BwPnLsaHJ/kEcCrdo6820wW2FwEfrKr/6tscBRyeZCNd79oL6cLm0YtRgyRJUgvGCW3fBvZYpM//IvB0YB1wHfAd4KXA2wbaHEUX0l5K9wSFc4FHVNWPF6kGSZKkJW+c0PZO4OVJ7lhVP9iaD6+qw+keUbWlNgW8pn9JkiStSOOEto/SPQz+s0mOpOstu5QR029U1SVbV54kSZJgvND2fbqAFuDNW2hXY+5fkiRJQ8YJVe9nlkltJUmStG2M80SEp26LQiRJkjS7sZ6IIEmSpMmaV2hL8rQk99nWxUiSJGm0+fa0HQ88dnBBkj9KctqiVyRJkqSb2Jrh0XXAfotUhyRJkrbAa9okSZIaYGiTJElqgKFNkiSpAQsJbU6oK0mSNCULmVz3iCRHDC9MsnmW9lVVPsZKkiRpESwkVGWB+15oe0mSJM1iXqGtqrz2TZIkaYoMY5IkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDVg9bQLkADWrT9p2iVIkrSk2dMmSZLUgKmGtiRPTPKxJD9MsinJeUmePKLds5JclOSqvs0B06hXkiRpWqbd0/ZCYBPwAuD/AKcD70/yFzMNkjwJeBvwXuBA4BvAiUnuNflyJUmSpmPa17QdUlWXDXx9WpLb04W5o/tlrwTeU1WvBkhyJvA/gfXAUydZrCRJ0rRMtadtKLDN+DJwW4AkdwbuBnxoYJvrgQ/T9bpJkiStCNMeHh1lX+Cb/Z/37t8vGGpzPnCbJGsnVpUkSdIULanQ1t9g8Bjgzf2iXfr3nw813Ti0XpIkaVlbMqEtyTrg/cC/V9XxQ6truPksy2f2dWiSc5Oc+9Of/nQxy5QkSZqKJRHaktwGOBn4Pje+uWCmR23noU1mvh7ugQOgqo6rqn2qap+1ax1BlSRJ7Zt6aEuyBjgR2B44qKquGFg9cy3b3kOb7Q38d1XZjSZJklaEaU+uu5ruTtC7AgdW1U8G11fVd4BvAU8c2OZm/dcnT7BUSZKkqZr2PG1vAR4NPJ/ubtD/PbDuy1V1NXAE8L4kG4DPAn9EF/L+YLKlSpIkTc+0Q9sj+/d/HLHuTsCGqvrnJLcGXgIcTvdEhIOr6usTqlGSJGnqphraqmrdPNu9HXj7tq1GkiRp6Zr6jQiSJEmam6FNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIasHraBUiS2rZu/UkT+6wNRx00sc+Slhp72iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAc7TJknSEOee01JkT5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgOmHtqS7JXk2CRfTbI5yRkj2iTJy5JcnORXSc5Kct8plCtJkjQVUw9twD2BRwPf6l+jrAcOB14HHAJsAk5NsvtEKpQkSZqypRDaTqiqParqicA3hlcmuQVdaDuyqo6pqlOBJwIFPHeypUqSJE3H1ENbVV0/R5N9gR2BDw1scwVwAnDgNixNkiRpyZh6aJuHvYHNwEVDy8/v10mSJC17LYS2XYBNVbV5aPlGYE2S7adQkyRJ0kStnnYB81QjlmW2dUkOBQ4F2HPPPbdhWZKkSVq3/qRplyBNTQs9bRuBHZKsGlq+M3BlVV07vEFVHVdV+1TVPmvXrp1IkZIkSdtSC6HtAmAVsNfQ8r37dZIkScteC6HtHOByumk+AEiyhm6+tpOnVZQkSdIkTf2atj6APbr/8g7Ajkme0H/98aq6MslRwOFJNtL1rr2QLnAePfGCJUmSpmDqoQ24LfDhoWUzX98J2AAcRRfSXgrsCpwLPKKqfjyhGiVJkqZq6qGtqjZww52gs7Up4DX9S5IkacVp4Zo2SZKkFc/QJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1YOpPRNDStW79SdMuQZIk9expkyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAasnnYBWph160+adgmSpEU0yZ/rG446aGKfpcVnT5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcB52iRJWiGcE65t9rRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG2SJEkNMLRJkiQ1wNAmSZLUAEObJElSAwxtkiRJDTC0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDVk+7gOVi3fqTpl2CJElLxqT+X9xw1EET+ZylwJ42SZKkBhjaJEmSGmBokyRJaoChTZIkqQGGNkmSpAYY2iRJkhpgaJMkSWqA87RJkqRmTXKe1GnPCWdPmyRJUgOaCW1J7pHk00muTHJJklclWTXtuiRJkiahieHRJLsApwLfBB4D3AV4A13oPGyKpUmSJE1EE6ENeDZwS+D3qupy4FNJdgSOSPK3/TJJkqRlq5Xh0QOBU4bC2Qfogtx+0ylJkiRpcloJbXsDFwwuqKrvA1f26yRJkpa1VkLbLsDPRyzf2K+TJEla1lq5pg2gRizLqOVJDgUO7b/clOTCbVnYhOwGXDbtIhri8Zo/j9X8eazmz2M1fx6rhZna8crrJvIxvznbilZC20Zg5xHLd2JED1xVHQcct62LmqQk51bVPtOuoxUer/nzWM2fx2r+PFbz57FamJV8vFoZHr2AoWvXkuwB3Iqha90kSZKWo1ZC28nAo5LsMLDs94FfAWdOpyRJkqTJaSW0vQ24GviXJA/vr1k7AnjjCpqjbVkN906Ax2v+PFbz57GaP4/V/HmsFmbFHq9Ujbq+f+lJcg/gGOC36a5jewdwRFVtnmphkiRJE9BMaJMkSVrJWhkeXbaS3CPJp5NcmeSSJK9Ksmoe2+2U5N1JNib5RZJ/SrLrJGqepnGOV5J1SWrE6wOTqnsakuyV5NgkX02yOckZ89xuxZ1b4xyrlXheJXliko8l+WGSTUnOS/LkeWx38yRvSPKTJFckOSnJum1f8XRtxfEadV59fhI1T0uSJyQ5J8nPklyV5MIkhyXZfo7tVtTPq1am/FiWkuwCnAp8E3gMcBfgDXRh+rA5Nv8gcHfgmcD1wOuAfwN+Z1vVO21bebwA/hL47MDXy31epHsCjwY+D2zxB9+QFXduMf6xgpV1Xr0Q+C7wArrv89HA+5PsVlVHb2G7NwFP6Lf7Kd01yZ9KckAkkOgAABBsSURBVO+qumrbljxV4x4v6H62fWTg619umxKXjF2B04G/o7sE6gF058nuwHO3sN3K+nlVVb6m9AJeSjcH3Y4Dy/6K7vFcO25hu9+mm1T4IQPLHtAve/i0v68leLzW9cfm4Gl/DxM+Xjcb+PNHgDPmsc1KPbfGOVYr7rwCdhux7P3Ad7ewzR2B64CnDSy7A3AN8Mxpf09L7Xj1bQp47rTrn/YLeA1dgMss61fczyuHR6frQOCUuvEdsB8AbgnsN8d2P66qs2YWVNUX6H6jO3BbFLpEjHu8VqSqun6MzVbkuTXmsVpxqmpUL+KXgdtuYbNH9u//MrCfHwJns4zPKRj7eOkGP2PLPd8r7ueVoW269mZocuCq+j5dz9HeI7eYZbve+XNs17pxj9eMd/fXK/0oyRuT3HJbFNm4lXpubY2Vfl7tS3fJwmz2Bn5QVZuGlq/Uc2qu4zXjiCTXJbksybuS3GZbF7YUJFmVZE2SBwPPA95afRfaCCvu55XXtE3XLox4DBfdEOAuY25350Woa6ka93hdDbwZ+CRwOfBQ4CV018Q9ZnFLbN5KPbfGseLPqyQH0H2vz9hCs3H/3S478zxeAO8BTqC7/m8f4HDgt5I8oJb/NFdXADfv//xe4MVbaLvifl4Z2qZv1G8QmWX5YmzXugV/31X1I258IesZSX4MvCXJfavqK4tcY+tW6rm1ICv9vOrv/nw/8O9VdfwczVf8ObWQ41VVTx/48qwk5wMfBw6hu8h+OdsXWEN3bdrL6eZn/bMttF9R55bDo9O1Edh5xPKdGP3bw1zb7TzHdq0b93iNMnNX1v22qqLlZ6WeW4tlRZxX/VDdycD3gafO0XzFn1MLPF6jfALYxDI/rwCq6ktVdXZVvZFuePQ5Se4yS/MVd24Z2qbrAobG3ZPsAdyK0eP0s27Xm218f7kY93iNUkPv6qzUc2uxLPvzKska4ES6C8QPqqor5tjkAmCPJLcaWr4izqkxjtdNDFzTtWzPq1l8qX+/0yzrV9zPK0PbdJ0MPCrJDgPLfh/4FXDmHNvt3l+oCUCSfejG8E/eFoUuEeMer1Ge0L+ftxiFLSMr9dxaLMv6vEqyGvgwcFfgwKr6yTw2+2T//riB/dyebh6tZX1OjXm8Ru3nd4Fbs0zPqy14UP/+3VnWr7ifVz7Gaor6yWK/CXydbkLAOwNvBP6hqg4baPdt4Myq+pOBZZ8A7kY3sefMhII/qarlOaEg4x+vJEcAO9BNgHo58BC6i1s/XlWPn+T3MEn9b/iP7r98EbAj8Ir+649X1ZWeW51xjtVKPK+SHAc8C3g+8IWh1V+uqquTfBqgqg4Y2O5Y4PHceHLdXYFlPbnuOMcryaF0Nx+cSjch7/3oJg+/ENh3ud6I0P/cORX4BrCZLrC9CDixqp7Ut/Hn1bQnilvpL+AewGl0vUU/Al4NrBpqswE4fmjZzsC76cbtL6e7wPUmEzkut9c4xwt4EnAu8Au6CT2/DbwKuPm0v59tfKzW0Q2njHqt89zaumO1Es+r/hjMdZzOYGhyYrq7Ad9IF9iuoLuo/k7T/n6W4vECDqD7ReBnwLXAxXRPlNhp2t/PNj5Wr6b7hXxT/7PnS8BfANsNHc/jh7ZbUT+v7GmTJElqgNe0SZIkNcDQJkmS1ABDmyRJUgMMbZIkSQ0wtEmSJDXA0CZJktQAQ5skSVIDDG3ShCWpodfmJJclOS3JU6ZU0y1G1DXX60kj9nP3gfUv28Ln3XrE/q5O8tMkX0xybJKHJ8kCv4+DB/b3jS20u02Sqwba7raQz1kqkuzX13/qLOsP7ddfn+S2I9bv3J9/G5PcrF928Ii/m2uSXJzkn5PcfwH1Df49X5vkN7bQ9osDbZ8wtO4jA+seMsv2r9/CefmgJB/sv4drkvwiybeT/FuSFyW5+Xy/J2maVk+7AGkFe2X/vh1wd+CxwMOS3L+qXjjhWq4bqGfQXwK3At5AN1P5oK+PaH9o/17AM5McWVuewfta4LX9n1cBuwD3Ap7R7+ucJE+pqg3z+SYGXAfcI8m+VXXOiPVPo5ul/zra/jn4OeBK4EFJbl5VVw+t35/u7yLAw4APDq1/KN0v76dX1fVD6y6im10eusd17Uv3FIgnJDm4qk5ZQJ0zx/mPueHv+9eS/Bbdo5vm8/fx+iQPnOO8Gtz3nwJvpTsOnwI+0q+6E/AA4DHAPwGXzmd/0lRN+5EMvnyttBf9Y2xGLD+A7tl519M/4mbaL7r/yArYfR5tt6d7TNFPgHf12z1ylra37tdvmmX9HYAT+jYXATvPs96D+21OoAsA756l3deB/wK+2rdv9rE3dA9kL+Bhs/z9nUQXuI8dsf5N/bZ/PuIYnjii/Rv6dV+bZ20zf88XAOf3xzwj2r25P+//rW//hKH1Hxk4Fwp48oh9vL5f96SBZTvTPfLuKrrndg5vE2A/4FbT/nv05Ws+L4dHpSWiqj5N959bgP81uC7JA/shoksHhqqOTXL74f0kOaMfJto+ycuTXNgPPR6/jb+F3wN2A94HvL1f9qxxdlRVPwQeB/wHsBddj99C/JDu+Zb/N8mOgyuS7AvcE3gH3X/yIyW5d5J/SvLD/pj/KMl7ktx5RNt7JPm7JF/qh7qvTvLdJG9JsvuI9jNDkH+Z5AFJTumH7DYlOXUhQ5DAp/v3/Yc+417A7ehC3WeH1w9t8+kR60Z5c/9+ryS3XECN0J0Td6b75WSwzlsCT6HrBdswxz5eQfec19fOc0jzfsAtgHNrRI9rdc6sqivmsS9p6gxt0tIycw3Xr8NEkj+m+0/3QOB04B/oHlT+TODcJHvOsq+PAn8GnNNv87VtVPOMmaHR46vqc8CFwGNGXUs1H1V1HXBk/+U41/q9HVgD/MHQ8mfR9cIdP9uGSR4HfBF4Ajccv7OAJwNfTHKPoU3+gG5I97t0ofUYugfIPxv4jyRrZ/moB/f7rb7eT9IFqTOS/OZ8vkngtP79gKHlBwysPx3Ya/BcSXI7uvB6SVVdMM/PGrzGcKEPrn4vXeB65tDy/wvsxA1Bf0u+Qxcc1wHPm0f7n/Xveya5xfzKlJawaXf1+fK10l7MPjz6cG4YHv3Nftnd6P6j+zZwh6H2+wObgX8dWn5G/xn/yVYO+zHP4VG63rDrgfMGlr203/YlI9pvcXh0oN1OM8cLWDuPemeG9t5Gd43cD4Zq2hG4YuaYAV9haHgU2J1uOPFHwF5D+78/3VDbZ4aW7wFsP6Kex/b7/7tZ6hw1FPiifvnfzvPv6GbARrrrA289sPzf6YaqAzyw3+fTB9Y/uV/23llqGzU8+sZ+3VfnWduvh0f7rz8AXD10vM/u69yOLhxvaXj0f9Nd9/jf/fd8m4E2o4ZHV9P9slJ0IfzZwG8B223Nvwtfvqb1sqdNmpIkR/Sv1yT5CPAJuv9g/6Gqvtc3ew7df2bPr27I8Neq6jTgY8AhSXYY8RGHV9Vl2/BbGHQoXe3HDyx7L12Qe+ZC7wKdUVW/oAtZALP1Vs227Wbg3cD9ktyvX/wUut63LfXq/AndzRfrq+rbQ/s8D/h/wIMHe62q6uKqumZEDf9G1/v2qFk+65Sq+sjQsuP69wdsocbBz7geOJMuoDwEIMmq/s9nVFUB5wG/5MZDpHMNjd5t4Bx9fZLPAS+g66X8q/nUNsLb6a59fFpf5/8AHkTXO3vtfHZQVRuB19Bdr/byOdpeRzfMfg7djQ5vpQvqm5Kck+SFSW415vciTVzLd01JrXtF/17Az4HPAO+sqvcNtPnt/n2/JDe6zq13W7oepbvR/cc86AvDjfvrsZ42tPj6qnrVAmsf3Od2wB/R9QjO3G1IVf0w3VQUj6S7c/G00XuY+yNmdjnGtu8EXkY3JPqc/v1iuoA8m5lj/r+S3GnE+nX9+/8Avg/QT5fxdOAPgXvTBYpVA9v89yyfde7wgqr6ZZJf0PUozden6e6C3J/uWr779TWc3u/zuiSfYWGh7a7ccI5eB/wY+BBdr+G5AEnuxk2Hn6+pqpvcIdo7je5mhGfS9drNDKm/Y47vb9jRwJ8Df5bk6Kr6r9ka9sH7QUnuTdebfX+6nsff7l/PSbJfVV2ywBqkiTO0SVNSVfPpfdq1f3/xHO1uPWLZqCkM7swN/xHP2AyMHdrohgBvC3y0qn42tO7ddKHtUMYIbUl2ousZg+7O1AWpqg19cPyDJB8A/ifwyrrp9BaDZo75n8+x+8FjfixdEPkBXWi6hG4YFbrvfUdG+/ksy6/jxqFvLjPB64Ch98Fjfgbw6CR372u7M3BhVf1gln2eVFUHz/G5d+Om59MVjJjWA7prApK8k+5Ggv3pQu6ZVfWtOT5neD/XJHkp3XDrUcAT57HN1xi4rrMPccfTBdy/BZ66kBqkaTC0SUvbL/r3narq8oVs2A+LDS87lRtfTL4YZnpLHp9ktt6wxyXZbYzh2of179/ZiqHe4+iC4/vohmvfOUf7mWN+l6r6zlw7T7KOLrB9Edivqn41tH6sO2gXoqq+meRS4LeS7ErXi3ZJVV040Oz0/n1/uuvKYP53jc72uSey8PPp3XS/JLyHLiAft+Xms372B5O8gG7euH3H2P5rSZ5BN1w66s5aacnxmjZpaft8//47U61iFv3w4QF0w3/vnOX1BbrrmP5ogfteTXczA3STn47rY3QXut8R+ERVXTxH+4Ue873695NHBLa7AjeZlmUbOY0uQD2K7jqx04fWf5kukO7Pwqf6WDRVdSndPHp3pDtv/mUrdjczFczrx9z+l/37Yv8iI20T9rRJS9sxdD1Zf5/kouFhpCTbAw+sqs9MpbruGrHQXYs38uL0JPehm8T2WXSTs86pn3/ubXQX41803+1GqaprkxxMN2HvqKc4DDuOLgy8NslXquqrQ7WtBh5cVWf0izb07w9Jkpkezn5od6xepDGdRnd92Xq6IeUbhbaq2pzkLLrey2voeh3PmGB9g15Ed6PKpVV11VyNZ1NVZyf5V7qbDW5yo0p/o8NDgfdV1S+H1t2M7npH6KZdkZY8Q5u0hFXVBf0QzruAbyT5BPAtujtK96TrDfopsPeka+vDyx/3X8465FhV/5nkXGCf/oLvMwdWb5/kiP7Pq+gunr8X3fxlq+mmg3hKfxfp2Krqi3TDl/Npe0m651d+EPhSkk/RzeYfuqk9HtTXtlvf/ttJTqSbKuO8JKcBt6Hr8bqMbsLkPbam/nma6TW7d/8+6hrC04FD+j+fV1Wz3SCxTVXVd+nuql0ML6E79nuNWLcr8BbgjUk+C3yDbjqX2wGPoPs39EO6oCsteYY2aYmrqvcl+Spd78TD6K7PuoLuYvePcNPnSU7KIXRzmn1m6NqpUd5ON+XCoXTTU8zYjhsuZL8GuBz4Hl1I/TDw6VHX5m1rVXVikvvSHfNH0PXWXEU3d9vJ3PD8yhl/QDf9xO8Bz6W70/LDdN/bRIYg+5suvkN3g8H3+mA0bLD3beJDo9tCVV2U5Fi64z7sPLqbFB5J12v7ZLpAvYnul593A2+aVniVFipT+HkoSZKkBfJGBEmSpAYY2iRJkhpgaJMkSWqAoU2SJKkBhjZJkqQGGNokSZIaYGiTJElqgKFNkiSpAYY2SZKkBhjaJEmSGvD/ARN7N1bj8/GtAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize = (10,10))\n",
    "x = np.array(means)\n",
    "hist = plt.hist(np.log10(x[x>0]), bins = 'auto')\n",
    "plt.tick_params(labelsize = 15)\n",
    "plt.xlabel(\"Per-TAD Mean WP-MNS\", size = 20)\n",
    "plt.ylabel(\"Frequency\", size = 20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}