version 2.3

Anwar-Said · Nov 8, 2023 · ecb8703 · ecb8703
1 parent 2dc52bb
commit ecb8703
Show file tree

Hide file tree

Showing 33 changed files with 597 additions and 427 deletions.
diff --git a/NeuroGraph/datasets.py b/NeuroGraph/datasets.py
@@ -4,19 +4,19 @@
 from typing import Callable, List, Optional
 
 import torch
-import zipfile
+
 from torch_geometric.data import (
     Data,
     InMemoryDataset,
     download_url,
-    extract_zip,
+    extract_zip
 )
 
-class NeuroGraphStatic(InMemoryDataset):
+class NeuroGraphDataset(InMemoryDataset):
     r"""The NeuroGraph benchmark datasets from the
     `"NeuroGraph: Benchmarks for Graph Machine Learning in Brain Connectomics"
     <https://arxiv.org/abs/2306.06202>`_ paper.
-    :class:`NeuroGraphStatic` holds a collection of five neuroimaging graph
+    :class:`NeuroGraphDataset` holds a collection of five neuroimaging graph
     learning datasets that span multiple categories of demographics, mental
     states, and cognitive traits.
     See the `documentation
@@ -26,22 +26,22 @@ class NeuroGraphStatic(InMemoryDataset):
     +--------------------+---------+----------------------+
     | Dataset            | #Graphs | Task                 |
     +====================+=========+======================+
-    | :obj:`HCP-State` | 7,443   | Graph Classification |
+    | :obj:`HCPActivity` | 7,443   | Graph Classification |
     +--------------------+---------+----------------------+
-    | :obj:`HCP-Gender`   | 1,078   | Graph Classification |
+    | :obj:`HCPGender`   | 1,078   | Graph Classification |
     +--------------------+---------+----------------------+
-    | :obj:`HCP-Age`      | 1,065   | Graph Classification |
+    | :obj:`HCPAge`      | 1,065   | Graph Classification |
     +--------------------+---------+----------------------+
-    | :obj:`HCP-FI`       | 1,071   | Graph Regression     |
+    | :obj:`HCPFI`       | 1,071   | Graph Regression     |
     +--------------------+---------+----------------------+
-    | :obj:`HCP-WM`       | 1,078   | Graph Regression     |
+    | :obj:`HCPWM`       | 1,078   | Graph Regression     |
     +--------------------+---------+----------------------+
 
     Args:
         root (str): Root directory where the dataset should be saved.
         name (str): The name of the dataset (one of :obj:`"HCPGender"`,
-            :obj:`"HCP-State"`, :obj:`"HCP-Age"`, :obj:`"HCP-FI"`,
-            :obj:`"HCP-WM"`).
+            :obj:`"HCPActivity"`, :obj:`"HCPAge"`, :obj:`"HCPFI"`,
+            :obj:`"HCPWM"`).
         transform (callable, optional): A function/transform that takes in an
             :obj:`torch_geometric.data.Data` object and returns a transformed
             version. The data object will be transformed before every access.
@@ -57,11 +57,11 @@ class NeuroGraphStatic(InMemoryDataset):
     """
     url = 'https://vanderbilt.box.com/shared/static'
     filenames = {
-        'HCP-Gender': 'r6hlz2arm7yiy6v6981cv2nzq3b0meax.zip',
-        'HCP-State': 'b4g59ibn8itegr0rpcd16m9ajb2qyddf.zip',
-        'HCP-Age': 'static/lzzks4472czy9f9vc8aikp7pdbknmtfe.zip',
-        'HCP-WM': 'xtmpa6712fidi94x6kevpsddf9skuoxy.zip',
-        'HCP-FI': 'g2md9h9snh7jh6eeay02k1kr9m4ido9f.zip',
+        'HCPGender': 'r6hlz2arm7yiy6v6981cv2nzq3b0meax.zip',
+        'HCPActivity': 'b4g59ibn8itegr0rpcd16m9ajb2qyddf.zip',
+        'HCPAge': 'static/lzzks4472czy9f9vc8aikp7pdbknmtfe.zip',
+        'HCPWM': 'xtmpa6712fidi94x6kevpsddf9skuoxy.zip',
+        'HCPFI': 'g2md9h9snh7jh6eeay02k1kr9m4ido9f.zip',
     }
 
     def __init__(
@@ -76,7 +76,7 @@ def __init__(
         self.name = name
 
         super().__init__(root, transform, pre_transform, pre_filter)
-        self.load(self.processed_paths[0])
+        self.data, self.slices = torch.load(self.processed_paths[0])
 
     @property
     def raw_dir(self) -> str:
@@ -124,8 +124,8 @@ def process(self):
                 sample = self.pre_transform(sample)
 
             data_list.append(sample)
-
-        self.save(data_list, self.processed_paths[0])
+        data, slices = self.collate(data_list)
+        torch.save((data,slices), self.processed_paths[0])
 
 class NeuroGraphDynamic():
     r"""Graph-based neuroimaging benchmark datasets, e.g.,
@@ -139,31 +139,56 @@ class NeuroGraphDynamic():
         Returns:
             list: A list of graphs in PyTorch Geometric (pyg) format. Each graph contains a list of dynamic graphs batched in pyg batch.
     """
+    url = 'https://vanderbilt.box.com/shared/static'
+    filenames = {
+            'DynHCPGender': 'mj0z6unea34lfz1hkdwsinj7g22yohxn.zip',
+            'DynHCPActivity': '2so3fnfqakeu6hktz322o3nm2c8ocus7.zip',
+            'DynHCPAge': '195f9teg4t4apn6kl6hbc4ib4g9addtq.zip',
+            'DynHCPWM': 'mxy8fq3ghm60q6h7uhnu80pgvfxs6xo2.zip',
+            'DynHCPFI': 'un7w3ohb2mmyjqt1ou2wm3g87y1lfuuo.zip',
+        }
     def __init__(self,root, name):
         self.root = root
         self.name = name
-        self.urls = {"DynHCP-Gender":'https://vanderbilt.box.com/shared/static/mj0z6unea34lfz1hkdwsinj7g22yohxn.zip',
-                    "DynHCP-State":'https://vanderbilt.box.com/shared/static/2so3fnfqakeu6hktz322o3nm2c8ocus7.zip',
-                    "DynHCP-Age":'https://vanderbilt.box.com/shared/static/195f9teg4t4apn6kl6hbc4ib4g9addtq.zip',
-                    "DynHCP-WM":'https://vanderbilt.box.com/shared/static/mxy8fq3ghm60q6h7uhnu80pgvfxs6xo2.zip',
-                    "DynHCP-FI":'https://vanderbilt.box.com/shared/static/un7w3ohb2mmyjqt1ou2wm3g87y1lfuuo.zip'
-                    }
-        if self.urls.get(name):
-            self.download(self.urls.get(name))
-        else:
-            print('dataset not found! The name of the datasets are: "DynHCPGender","DynHCPActivity","DynHCPAge","DynHCPWM","DynHCPFI"')
+
+
+        # self.urls = {"DynHCP-Gender":'https://vanderbilt.box.com/shared/static/mj0z6unea34lfz1hkdwsinj7g22yohxn.zip',
+        #             "DynHCP-State":'https://vanderbilt.box.com/shared/static/2so3fnfqakeu6hktz322o3nm2c8ocus7.zip',
+        #             "DynHCP-Age":'https://vanderbilt.box.com/shared/static/195f9teg4t4apn6kl6hbc4ib4g9addtq.zip',
+        #             "DynHCP-WM":'https://vanderbilt.box.com/shared/static/mxy8fq3ghm60q6h7uhnu80pgvfxs6xo2.zip',
+        #             "DynHCP-FI":'https://vanderbilt.box.com/shared/static/un7w3ohb2mmyjqt1ou2wm3g87y1lfuuo.zip'
+        #             }
+        assert name in self.filenames.keys()
+        self.name = name
+        file_path = os.path.join(self.root,self.name,'processed', self.name+".pt")
+        if not os.path.exists(file_path):
+            self.download()
+        # else:
+        #     print('dataset not found! The name of the datasets are: "DynHCP-Gender","DynHCP-Activity","DynHCP-Age","DynHCP-WM","DynHCP-FI"')
         self.dataset, self.labels = self.load_data()
 
-    def download(self,url):
-
-        download_url(url, os.path.join(self.root, self.name))
-        basename = os.path.basename(url)
-        with zipfile.ZipFile(os.path.join(self.root,self.name,basename), 'r') as file:
-            file.extractall(os.path.join(self.root,self.name,os.path.dirname(basename)))
+    # def download(self,url):
+    #     download_url(url, os.path.join(self.root, self.name))
+    #     basename = os.path.basename(url)
+    #     with zipfile.ZipFile(os.path.join(self.root,self.name,basename), 'r') as file:
+    #         file.extractall(os.path.join(self.root,self.name,os.path.dirname(basename)))
             # self.remove(os.path.join(self.raw_dir,basename))
+
+
+    def download(self):
+        url = f'{self.url}/{self.filenames[self.name]}'
+        path = download_url(url, os.path.join(self.root, self.name))
+        extract_zip(path, self.root)
+        os.unlink(path)
+        # os.rename(
+        #     osp.join(self.raw_dir, self.name, 'processed', f'{self.name}.pt'),
+        #     osp.join(self.raw_dir, 'data.pt'))
+        # shutil.rmtree(osp.join(self.raw_dir, self.name))
+
+
     def load_data(self):
-        if self.name=='DynHCP-State':
-            dataset_raw = torch.load(os.path.join(self.root,self.name,self.name,"processed", self.name+".pt"))
+        if self.name=='DynHCPActivity':
+            dataset_raw = torch.load(os.path.join(self.root,self.name,'processed', self.name+".pt"))
             dataset,labels = [],[]
             for v in dataset_raw:
                 batches = v.get('batches')
@@ -173,7 +198,7 @@ def load_data(self):
                         dataset.append(b)
                         labels.append(y)
         else:
-            dataset = torch.load(os.path.join(self.root,self.name,self.name,"processed", self.name+".pt"))
+            dataset = torch.load(os.path.join(self.root,self.name,'processed', self.name+".pt"))
             labels = dataset['labels']
             dataset = dataset['batches']
         return dataset,labels

diff --git a/doc/_build/doctrees/NeuroGraph.doctree b/doc/_build/doctrees/NeuroGraph.doctree
diff --git a/doc/_build/doctrees/datasets.doctree b/doc/_build/doctrees/datasets.doctree
diff --git a/doc/_build/doctrees/environment.pickle b/doc/_build/doctrees/environment.pickle
diff --git a/doc/_build/doctrees/get_started.doctree b/doc/_build/doctrees/get_started.doctree
diff --git a/doc/_build/doctrees/index.doctree b/doc/_build/doctrees/index.doctree
diff --git a/doc/_build/doctrees/install.doctree b/doc/_build/doctrees/install.doctree
diff --git a/doc/_build/doctrees/preprocess.doctree b/doc/_build/doctrees/preprocess.doctree
diff --git a/doc/_build/doctrees/utils.doctree b/doc/_build/doctrees/utils.doctree
diff --git a/doc/_build/html/.buildinfo b/doc/_build/html/.buildinfo
@@ -1,4 +1,4 @@
 # Sphinx build info version 1
 # This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: 7313becfcf9696e10c17e4e0753ff7df
+config: c0e52ab7b160affdaa25d3a6a797e2b5
 tags: 645f666f9bcd5a90fca523b33c5a78b7
diff --git a/doc/_build/html/NeuroGraph.html b/doc/_build/html/NeuroGraph.html
@@ -11,11 +11,9 @@
     <script src="_static/js/html5shiv.min.js"></script>
   <![endif]-->
 
-        <script data-url_root="./" id="documentation_options" src="_static/documentation_options.js"></script>
-        <script src="_static/jquery.js"></script>
-        <script src="_static/underscore.js"></script>
-        <script src="_static/_sphinx_javascript_frameworks_compat.js"></script>
-        <script src="_static/doctools.js"></script>
+        <script src="_static/documentation_options.js?v=20623aea"></script>
+        <script src="_static/doctools.js?v=888ff710"></script>
+        <script src="_static/sphinx_highlight.js?v=dc90522c"></script>
     <script src="_static/js/theme.js"></script>
     <link rel="index" title="Index" href="genindex.html" />
     <link rel="search" title="Search" href="search.html" />
@@ -97,16 +95,16 @@
 
   <p>NeuroGraph is a collection of graph-based neuroimaging datasets that span multiple categories of demographics, mental states and cognitive traits. The following provides an overview of these categories and their associated datasets.</p>
 <section id="demographics">
-<h1>Demographics<a class="headerlink" href="#demographics" title="Permalink to this heading">¶</a></h1>
+<h1>Demographics<a class="headerlink" href="#demographics" title="Link to this heading">¶</a></h1>
 <p>Demographics category includes gender and age estimation. The gender attribute facilitates a binary classification with the categories being male and female. Age is categorized into three distinct groups as in: 22-25, 26-30, and 31-35 years.  We introduce four datasets named: HCP-Gender, HCP-Age, DynHCP-Gender, and DynHCP-Age under this category. The first two are
 static graph datasets while the last two are the corresponding dynamic graph datasets.</p>
 </section>
 <section id="mental-states">
-<h1>Mental States<a class="headerlink" href="#mental-states" title="Permalink to this heading">¶</a></h1>
+<h1>Mental States<a class="headerlink" href="#mental-states" title="Link to this heading">¶</a></h1>
 <p>The mental state decoding involves seven tasks: Emotion Processing, Gambling, Language, Motor, Relational Processing, Social Cognition, and Working Memory. Each task is designed to help delineate a core set of functions relevant to different facets of the relation between human brain, cognition and behavior. Under this category, we present two datasets: HCP-Activity, a static representation, and DynHCP-Activity, its dynamic counterpart.</p>
 </section>
 <section id="cognitive-traits">
-<h1>Cognitive Traits<a class="headerlink" href="#cognitive-traits" title="Permalink to this heading">¶</a></h1>
+<h1>Cognitive Traits<a class="headerlink" href="#cognitive-traits" title="Link to this heading">¶</a></h1>
 <p>The cognitive traits category of our dataset comprises two significant traits: working memory (List Sorting) and fluid intelligence evaluation with PMAT24. Working memory refers to an individual’s capacity to temporarily hold and manipulate information, a crucial aspect that influences higher cognitive functions such as reasoning, comprehension, and learning. Fluid intelligence represents the ability to solve novel problems, independent of any knowledge from the past. It demonstrates the capacity to analyze complex relationships, identify patterns, and derive solutions in dynamic situations. The prediction of both these traits, quantified as continuous variables in our dataset, are treated as regression problem. We aim to predict
 the performance or scores related to these cognitive traits based on the functional connectome graphs. We generate four datasets under cognitive traits: HCP Fluid Intelligence (HCP-FI), HCP Working Memory (HCP-WM), DynHCP-FI and DynHCP-WM.</p>
 </section>