Make a taxonomy file compatible with mothur

make_mothur_tax.py

#!/usr/bin/python
"""Make a taxonomy file compatible with mothur."""

import os
import sys
import pandas as pd

if len(sys.argv) != 4:
    print("Please specify the seq_info.csv, tax_info.csv, and output.tsv files")

# Get the filepaths
seqinfo = sys.argv[1]
assert os.path.exists(seqinfo)

taxinfo = sys.argv[2]
assert os.path.exists(taxinfo)

output = sys.argv[3]

# Read in the tables
taxinfo = pd.read_table(taxinfo, sep=',')
assert "tax_id" in taxinfo
taxinfo.set_index("tax_id", inplace=True)

seqinfo = pd.read_table(seqinfo, sep=',')


# Function to generate the taxonomy string for a certain taxid
def taxonomy_string(taxid, taxinfo, sep=';', root=1):
    tax_string = []
    while taxid in taxinfo.index.values:
        tax_string.append(taxinfo.loc[taxid, "tax_name"])
        if taxid == root:
            break
        if taxinfo.loc[taxid, "parent_id"] == taxid:
            break

        taxid = taxinfo.loc[taxid, "parent_id"]
    assert tax_string[-1] == taxinfo.loc[root, "tax_name"]
    tax_string = tax_string[::-1]
    return sep.join(tax_string) + sep

# Write a table with the sequence name and taxonomy string for each sequence
cache = {}
with open(output, "wt") as fo:
    for ix, r in seqinfo.iterrows():
        if r["tax_id"] not in cache:
            cache[r["tax_id"]] = taxonomy_string(r["tax_id"], taxinfo)
        fo.write("{}\t{}\n".format(r["seqname"], cache[r["tax_id"]]))