This script extracts insert size mean and std, mass of mapped read information from Cufflinks log outputs.

getinsertsize.py

#!/bin/env python3

import os
import re
import sys

if len(sys.argv)<2:
        print('This script extracts insert size information from Cufflinks logs.',file=sys.stderr);
        print('Usage: getinsertsize [cufflinks log file]',file=sys.stderr);
        print('Note: you may specify different log files using filename wildcards.',file=sys.stderr);
        sys.exit();


flist=sys.argv[1:];


print('File\tMapMass\tReadLength\tMean\tStd');

for fof in sorted(flist):
        jobid=re.findall('e(\d+)',fof)[0];
        ispassed=False;
        nline=0;
        mpms='';
        rtype='';
        dmean='';
        dstd='';
        for lines in open(fof):
                if lines.startswith('>  '):
                        r1=re.findall('Total Map Mass: ([\d\.]+)',lines);
                        r2=re.findall('Read Type: ([\d]+)bp',lines);
                        r3=re.findall('Estimated Mean: ([\d\.]+)',lines);
                        r4=re.findall('Estimated Std Dev: ([\d\.]+)',lines);
                        if len(r1)>0: mpms=r1[0];
                        if len(r2)>0: rtype=r2[0];
                        if len(r3)>0: dmean=r3[0];
                        if len(r4)>0: dstd=r4[0];
                        ispassed=True;
                else:
                        if ispassed: break;
        print(fof+'\t'+str(mpms)+'\t'+str(rtype)+'\t'+str(dmean)+'\t'+str(dstd));