comparison comp_clono_CDR3.py @ 12:cdf95051bc55 draft default tip

Uploaded 2 tools
author chmaramis
date Sun, 18 Mar 2018 07:11:06 -0400
parents
children
comparison
equal deleted inserted replaced
11:14896ea6e180 12:cdf95051bc55
1 # -*- coding: utf-8 -*-
2 """
3 Created on Thu Jun 19 17:33:34 2014
4
5 @author: chmaramis
6 """
7
8 from __future__ import division
9 import numpy as np
10 from pandas import *
11 import functools as ft
12 import sys
13 import time
14
15 frm = lambda x,y: '{r}/{l}'.format(r=x,l=y)
16
17
18 def cdr3Computation(inp_name, out1, t10n, fname):
19
20 frame = DataFrame()
21 tp = read_csv(inp_name, iterator=True, chunksize=5000,sep='\t', index_col=0 )
22 frame = concat([chunk for chunk in tp])
23
24 grouped = frame.groupby(['AA JUNCTION'])
25 x=grouped.size()
26 x1=DataFrame(x.index, columns=['AA JUNCTION'])
27 x1['Reads']=x.values
28 total = sum(x1['Reads'])
29 #x1['Reads/Total'] = ['{r}/{l}'.format(r=pr , l = total) for pr in x1['Reads']]
30 x1['Reads/Total'] = x1['Reads'].map(ft.partial(frm, y=total))
31 x1['Frequency %'] = (100*x1['Reads']/total).map('{:.4f}'.format)
32
33 final = x1.sort_values(by = ['Reads'] , ascending = False)
34
35 final.index=range(1,len(final)+1)
36 final.to_csv(out1 , sep = '\t')
37
38 numofclono = len(final)
39 clust = len(final[final['Reads'] > 1])
40 sing = len (final[final['Reads'] == 1])
41 top10 = final[['AA JUNCTION','Frequency %']].head(10)
42 top10.to_csv(t10n , sep = '\t')
43
44 summary = [[str(top10['AA JUNCTION'].values[0])]]
45 summary.append([top10['Frequency %'].values[0]])
46 summary.append([numofclono])
47 summary.append([clust,'{:.4f}'.format(100*clust/numofclono)])
48 summary.append([sing,'{:.4f}'.format(100*sing/numofclono)])
49
50 ind = ['Dominant Clonotype (CDR3)', 'Frequency', 'Number of Clonotypes' , 'Expanding Clonotypes','Singletons']
51 spl = fname.split('_')
52 col = [spl[0],'%']
53
54 frsum = DataFrame(summary,index = ind, columns = col)
55
56 return frsum
57
58 if __name__ == '__main__':
59
60 start=time.time()
61
62 # Parse input arguments
63 inp_name = sys.argv[1]
64 out1 = sys.argv[2]
65 t10n = sys.argv[3]
66 sname = sys.argv[4]
67 fname = sys.argv[5]
68
69 # Execute basic function
70 frsum = cdr3Computation(inp_name,out1,t10n,fname)
71
72 # Save output to CSV files
73 if not frsum.empty:
74 frsum.to_csv(sname, sep = '\t')
75
76 # Print execution time
77 stop=time.time()
78 print('Runtime:' + str(stop-start))