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