Using sj2psi

gistfile1.txt

import sj2psi
from glob import iglob

stepsize = 100

def i_to_range(i, stepsize):
    if i % stepsize == 0:
        return ((i/stepsize)-1)*stepsize + 1, i
    else:
        return (i/stepsize)*stepsize + 1, i

def filter_and_write(i, dfs, stepsize=stepsize):
    summary = pd.concat(dfs)
    csv = 'sj_{}-{}.csv'.format(*i_to_range(i, stepsize))
    summary.to_csv(csv)
    print 'wrote {}'.format(csv)


dfs = []
for i, filename in enumerate(iglob('/home/obotvinnik/scratch/mn_diff_singlecell/fastq/sj_out_tab/*SJ.out.tab')):
    sample_id = os.path.basename(filename).split('.')[0]
    if i % 10 == 0:
        print i, 'on sample id {}'.format(sample_id)
    
    sj = sj2psi.read_sj_out_tab(filename)
    sj['sample_id'] = sample_id
    sj = sj2psi.get_psis(sj)
    dfs.append(sj)
    if i % stepsize == 0 and i > 0:
        filter_and_write(i, dfs)
        dfs = []
filter_and_write(i, dfs)