Merge pull request #44 from zm711/merge-datasets

Add ability to merge datasets for plotting

docs/source/API.rst

@@ -24,4 +24,9 @@ spikeanalysis
     .. autoclass:: AnalogAnalysis
         :members:
 
+    .. autoclass:: MergedSpikeAnalysis
+        :members:
+
     .. autofunction:: kolmo_smir_stats
+
+    .. autofunction:: prevalence_counts

docs/source/conf.py

@@ -24,7 +24,7 @@
 author = "Zach McKenzie"
 
 # The full version, including alpha/beta/rc tags
-release = "0.0.11"
+release = "0.1.0"
 
 
 # -- General configuration ---------------------------------------------------

docs/source/submodules/curated_spike_analysis.rst

@@ -61,3 +61,17 @@ To revert back to the original full set of neurons use :code:`revert_curation()`
     curated_st.revert_curation()
 
 
+Plotting the Data
+-----------------
+
+Since :code:`CuratedSpikeAnalysis` inherits from :code:`SpikeAnalysis` it can be used with 
+the :code:`SpikePlotter` class with no additional work.
+
+.. code-block:: python
+
+    plotter = sa.SpikePlotter()
+
+    plotter.set_analysis(curated_st)
+    plotter.plot_zscores()
+
+

docs/source/submodules/index.rst

@@ -13,3 +13,4 @@ Submodules
    analog_analysis
    curated_spike_analysis
    functions
+   merged_spike_analysis

docs/source/submodules/merged_spike_analysis.rst

@@ -0,0 +1,52 @@
+MergedSpikeAnalysis
+===================
+
+Module for merging datasets. Once data has been curated it may be beneficial to look at a series of 
+animals altogether. To facilitate this the MergedSpikeAnalysis object can be used. This is done in
+similar fashion to other classes
+
+.. code-block:: python
+
+    import spikeanalysis as sa
+
+    # we start with SpikeAnalysis or CuratedSpikeAnalysis objects st1
+    # and st2
+    merged_data = sa.MergedSpikeAnalysis(spikeanalysis_list=[st1, st2], name_list=['animal1', 'animal2'])
+
+    # if we need to add an animal, st3 we can use
+    merged_data.add_analysis(analysis=st3, name='animal3')
+
+    # or we can use lists
+    merged_data.add_analysis(analysis=[st3,st4], name=['animal3', 'animal4'])
+
+Once the data to merge is ready to be merged one can use the :code:`merge()` function. This takes
+in the value :code:`psth`, which can either be set to :code:`True` to mean to load a balanced 
+:code:`psths` values or can be a value in a list of potential merge values, e.g. :code:`zscore` or
+for example :code:`fr`.
+
+.. code-block:: python
+
+    # will attempt to merge the psths of each dataset
+    merged_data.merge(psth=True)
+
+    # will attempt to merge z scores
+    merged_data.merge(psth=['zscore'])
+
+Note, that the datasets to be merged must be balanced. For example a dataset with 5 neurons,
+10 trials, and 200 timepoints can only be merged to another dataset with :code:`x` neurons, 10 
+trials, and 200 timepoints. The concatenation occurs at the level of the neuron axis (:code:`axis 0`)
+so everything else must have the same dimensionality.
+
+Finally, the merged data set can be return for use in the :code:`SpikePlotter` class.
+
+.. code-block:: python
+
+    msa = merged_data.get_merged_data()
+    plotter = sa.SpikePlotter()
+    plotter.set_analysis(msa)
+
+This works because the :code:`MSA` returned is a :code:`SpikeAnalysis` object that has specific
+guardrails around methods which can no longer be accessed. For example, if the data was merged with
+:code:`psth=True`, then z scores can be regenerated across the data with a different :code:`time_bin_ms`,
+but if :code:`psth=['zscore']` was used then new z scores can be generated and the :code:`MSA` will
+return a :code:`NotImplementedError`

docs/source/submodules/spike_data.rst

@@ -190,6 +190,20 @@ as understandable and maintainable a weighted average is used (faster, slightly
     \frac{1}{\Sigma amplitudes} \sum{amplitudes * depths}
     
 
+Waveform amplitudes
+^^^^^^^^^^^^^^^^^^^
+
+Since neurons should always have the same amplitude we can assess the variation in amplitudes
+as a measure of the quality of a neuron. We expect a rather gaussian distribution of amplitudes
+so using the :code:`get_amplitudes()` function we assess how many spikes fall within a certain
+std of the waveform data.
+
+.. code-block:: python
+
+    spikes.get_amplitudes(std=2) # 2 std devs
+
+
+
 Pipeline Function
 -----------------
 
@@ -204,6 +218,8 @@ parameters to be provided. Example below will all values included.
         idthres=20, # isolation distance 20--need an empiric number from your data
         rpv=0.02, # 2% the amount of spikes violating the 2ms refractory period allowed
         sil=0.45, # silhouette score (-1,1) with values above 0 indicates better and better clustering
+        amp_std=2 # number of std deviations above mean waveform amplitude to look at
+        amp_cutoff=0.98, # percent of neurons which must fall within amp_std deviations of mean waveform
         recurated= False, # I haven't recurated my data
         set_caching = True, # I want to save data for future use
         depth= 500, # probe inserted 500 um deep

docs/source/submodules/stimulus_data.rst

@@ -96,7 +96,7 @@ they can be returned using :code:`get_stimulus_channels`. Finally stimulus' shou
 
     stim_dict = stim.get_stimulus_channels()
     stim.set_trial_groups(trial_dictionary=trial_dictionary) # dict as explained above
-    sitm.set_stimulus_names(stim_names = name_dictionary) # same keys with str values
+    sitm.set_stimulus_name(stim_names = name_dictionary) # same keys with str values
 
 
 Train-based data
@@ -143,7 +143,6 @@ generated stimulus data. To load it simply requires:
     stim.get_all_files()
 
 
-
 Convenience Pipeline
 --------------------
 

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "spikeanalysis"
-version = '0.0.11'
+version = '0.1.0'
 authors = [{name="Zach McKenzie", email="[email protected]"}]
 description = "Analysis of Spike Trains"
 requires-python = ">=3.9"

src/spikeanalysis/__init__.py

@@ -5,8 +5,10 @@
 from .intrinsic_plotter import IntrinsicPlotter
 from .analog_analysis import AnalogAnalysis
 from .curated_spike_analysis import CuratedSpikeAnalysis, read_responsive_neurons
+from .merged_spike_analysis import MergedSpikeAnalysis
 from .stats_functions import kolmo_smir_stats
 from .plotting_functions import plot_piechart
+from .utils import prevalence_counts
 
 import importlib.metadata
 

src/spikeanalysis/analysis_utils/histogram_functions.py

@@ -1,3 +1,4 @@
+from __future__ import annotations
 import numpy as np
 from numba import jit
 import numba

src/spikeanalysis/analysis_utils/latency_functions.py

@@ -1,3 +1,4 @@
+from __future__ import annotations
 import numpy as np
 from numba import jit
 import math