annotate cmpb2016/comp_clono_VDJCDR3.py @ 1:1319c163872c draft

Uploaded README file

# -*- coding: utf-8 -*-
"""
Created on Thu Dec  3 14:54:00 2015

@author: chmaramis
"""

from __future__ import division
import numpy as np
from pandas import *
import functools as ft
import sys
import time

frm = lambda x,y: '{r}/{l}'.format(r=x,l=y)

def clonotypeComputationVDJ(inp_name,out1,t10n,fname):

    frame = DataFrame()
    tp = read_csv(inp_name, iterator=True, chunksize=5000,sep='\t', index_col=0 )
    frame = concat([chunk for chunk in tp]) 

    grouped = frame.groupby(['V-GENE','D-GENE','J-GENE','AA JUNCTION'])
    x=grouped.size()
    x1=DataFrame(list(x.index), columns=['V-GENE','D-GENE','J-GENE','AA JUNCTION'])
    x1['Reads']=x.values
    total = sum(x1['Reads'])
    #x1['Reads/Total'] = ['{r}/{l}'.format(r=pr , l = total) for pr in x1['Reads']]
    x1['Reads/Total'] = x1['Reads'].map(ft.partial(frm, y=total))
    x1['Frequency %'] = (100*x1['Reads']/total).map('{:.4f}'.format)

    final = x1.sort_values(by = ['Reads'] , ascending = False)
    #final = x1.sort_values(by = ['Reads'] , ascending = False)
    
    final.index=range(1,len(final)+1)
    final.to_csv(out1 , sep = '\t')
    
    numofclono = len(final)
    clust = len(final[final['Reads'] > 1])
    sing = len (final[final['Reads'] == 1])
    top10 = final[['V-GENE','D-GENE','J-GENE','AA JUNCTION','Frequency %']].head(10)
    top10.to_csv(t10n , sep = '\t')
    
    summary = [[str(top10['V-GENE'].values[0]+','+top10['D-GENE'].values[0]+','+top10['J-GENE'].values[0]+','+top10['AA JUNCTION'].values[0])]]
    summary.append([top10['Frequency %'].values[0]])
    summary.append([numofclono])
    summary.append([clust,'{:.4f}'.format(100*clust/numofclono)])
    summary.append([sing,'{:.4f}'.format(100*sing/numofclono)])
    
    
    ind = ['Dominant Clonotype (V+D+J+CDR3)', 'Frequency', 'Number of Clonotypes' , 'Expanding Clonotypes', 'Singletons']
    spl = fname.split('_')
    col = [spl[0],'%']
    
    frsum = DataFrame(summary,index = ind, columns = col)
    
    return frsum

if __name__ == '__main__':   

    start=time.time()

    # Parse input arguments    
    inp_name = sys.argv[1]
    out1 = sys.argv[2]
    t10n = sys.argv[3]
    sname = sys.argv[4]
    fname = sys.argv[5]
            
    # Execute basic function
    frsum = clonotypeComputationVDJ(inp_name,out1,t10n,fname)
    
    # Save output to CSV files
    if not frsum.empty: 
        frsum.to_csv(sname, sep = '\t')
        
    # Print execution time
    stop=time.time()
    print('Runtime:' + str(stop-start))