Count Hits to Each Unique Domain in Hmmscan Results w/ Python?

Question

[ethanabaker1]

I'm trying to get some info about the top domains that my data matched to using hmmscan...I'm using the domain table output that is packaged with hmmscan (I also have the PFAM file). I want to use some sort of script to go through and tell me how many matches to each unique domain there are and then sort that list so that I can extract the top 15-20 domains that my data matched to. Thoughts?

Answer 1

[5heikki]

Given that domain info is in column X

cut -f X yourOutputFile | sort | uniq -c | sort -k1,1g

Answer 2

[John]

One option is to load your hmmscan table results as a pandas.DataFrame, then counting the domains is easy with the value_counts() method:

# Here, `hmm_tbl` is a pandas.DataFrame with your hmmscan results.
# This dataframe has a column `accession_target`, which is the accession of the result in the table.
# Eg, the dataframe might look something like this (truncated for readability)
#     acc  accession_query accession_target  ali_coord_from  env_coord_from
# 0  0.91              NaN       PF00005.25              10               6
# 1  0.89              NaN        PF13304.4              33              21
hmm_tbl.accession_target.value_counts()
# PF07719.15    4668
# PF00005.25    4402
# PF13304.4     3626
# PF13432.4     3601
# PF13428.4     3513
# PF00515.26    3494
# ...

Link: pandas.Series.value_counts — pandas 0.21.0 documentation