Getting simplified regulons list

[topherconley]

I used the pyscenic command line interface (CLI) and it is a bit limited in that it won't return a simple list of pruned regulons with their targets. I believe this is important for interpreting the AUC matrix of activity scores and knowing what the targets are of the transcription factors. I cobbled together some python code to get this list for downstream analysis (see below). If you could make this standard output from the `pyscenic ctx' output that would be helpful. The current 'CTX' pruned file has a non-standard means of parsing and appears to have a lot of information that is of secondary interest.

import pandas as pd
import os
from pyscenic.transform import df2regulons as df2regs
import json

# use function from pyscenic for parsing the 'pyscenic ctx' output
def _df2regulons(fname):
  ext = os.path.splitext(fname,)[1]
  df = pd.read_csv(fname, sep=',', index_col=[0,1], header=[0,1], skipinitialspace=True)
  df[('Enrichment', 'Context')] = df[('Enrichment', 'Context')].apply(lambda s: eval(s))
  df[('Enrichment', 'TargetGenes')] = df[('Enrichment', 'TargetGenes')].apply(lambda s: eval(s))
  return df2regs(df)

# Form dictionary of { TF : Target } pairs from 'pyscenic ctx' output.
rdict = {}
regulons = _df2regulons('joint_ctx.csv')
for reg in regulons:
  targets = [ target for target in reg.gene2weight ]
  rdict[reg.name] = targets

# Write to JSON for import to R
rjson = json.dumps(rdict)
f = open("regulons.json","w")
f.write(rjson)
f.close()

Then export into R with :

library(jsonlite)
regs <- read_json('regulons.json', simplify = T)

[bramvds]

Dear Chris,

Thanks for the suggestion. I'll make JSON export of regulons available as an option in the CLI version of pySCENIC in the next release.

Kr, Bram

[bramvds]

Dear Chris,

Just released a new version (0.8.7). This version includes your suggestion. Thanks.

Kindest regards, Bram

[topherconley]

Thanks!

[finjen]

Hi, I am trying to export regulons using the above mentioned code, however, I obtain the following error:

TypeError: expected str, bytes or os.PathLike object, not DataFrame

Could anyone please let me know what might be going wrong here? Thank you.

Best.