Skip to main content

Unsupervised clustering algorithm for 2D data

Project description

rastermap

This algorithm computes a 2D continuous sorting of neural activity. It assumes that the spike matrix S is neurons by timepoints.

Here is what the output looks like for a segment of a mesoscope recording (2.5Hz sampling rate):

rastersorted

Here is an example using the algorithm (also see this jupyter-notebook)

from rastermap import rastermap # <-- if pip installed
import numpy as np
from scipy.ndimage import gaussian_filter1d
import matplotlib.pyplot as plt

# load data (S is neurons x time)
# these files are the outputs of suite2p
S = np.load('spks.npy')
iscell = np.load('iscell.npy')
S = S[iscell[:,0].astype(bool),:]

# run rastermap 
# (will take ~ 30s for 6000 neurons x 20000 timepts on a laptop)
# these are the default options, you can change them and input them to the function
ops = {'nclust': 30, # number of clusters
       'iPC': np.arange(0,200).astype(np.int32), # number of PCs to use for mapping algorithm
       'upsamp': 100, # upsampling factor for embedding position
       'sigUp': 1, # standard deviation for kriging upsampling
       'equal': False # whether or not clusters should be of equal size (recommended False!)
       }
# user options
isort1,isort2 = rastermap.main(S,ops)

# if you just want to use the defaults
isort1,isort2 = rastermap.main(S)

# sort neurons and smooth across neurons and zscore in time
# smoothing will take ~ 10s depending on data size
Sm = gaussian_filter1d(S[isort1,:].T, np.minimum(10,int(S.shape[0]*0.005)), axis=1)
Sm = Sm.T
Sm = zscore(Sm, axis=1)

# (optional) smooth in time
Sm = gaussian_filter1d(Sm, 1, axis=1)

# view neuron sorting :)
fs = 2.5 # sampling rate of data in Hz
sp = Sm[:,1000:3000]
plt.figure(figsize=(16,12))
ax=plt.imshow(sp,vmin=0,vmax=3,aspect='auto',extent=[0,sp.shape[1]/fs, 0,sp.shape[0]])
plt.xlabel('time (s)', fontsize=18)
plt.ylabel('neurons', fontsize=18)
plt.show()

If you don't pip install the package, you can also run it using the path to this github folder

import sys
sys.path.insert(0, '/media/carsen/DATA2/github/rastermap/rastermap/')
import rastermap

Installation

You can just download the github folder as outlined above or you can pip install the package:

pip install rastermap

Requirements

This package was written for Python 3 and relies on numpy and scipy. The Python3.x Anaconda distributions will contain all the dependencies.

Matlab

The matlab code needs to be cleaned up but the main function to call is mapTmap.m. This function is used in the example script loadFromPython.m.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rastermap-0.0.3.tar.gz (618.4 kB view hashes)

Uploaded Source

Built Distribution

rastermap-0.0.3-py3-none-any.whl (604.7 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page