Fix the T2T gff3 file to prevent errors when getting a gffutils db

fix_t2t_gff.py

#!/usr/bin/env python

# fix t2t gff3 error
# now the problem is ids for some start&stop condons/exons are not unique
# 11/20/2022

import sys,gzip
from collections import defaultdict

def sep_info(info):
    items = info.split(';')
    items_tuple = [(i.split('=')[0], i.split('=')[1]) for i in items]
    return items_tuple

def comb_info(items):
    res = [f'{i[0]}={i[1]}' for i in items]
    return ';'.join(res)

def update_id(r, info, count, replace=False):
    (idx, sid) = [(i, j) for i,j in enumerate(info) if j[0] == 'ID'][0]
    if replace:
        tmp = sid[1].split(':')
        new_id = f'{":".join(tmp[:-1])}:{count}'
    else:
        new_id = f'{sid[1]}:{count}'
    info[idx] = (sid[0], new_id)
    r[8] = comb_info(info)
    return r

def fix_id(gff3, out):
    start_codon_cnt = defaultdict(lambda:0)
    stop_codon_cnt = defaultdict(lambda:0)
    liftoff_id_cnt = defaultdict(lambda:0)

    with open(gff3, 'r') as f, gzip.open(out, 'wb') as out:
        for row in f:
            if row.startswith('#'): 
                out.write(row.encode())
                continue

            r = row.rstrip().split('\t')
            if r[2] == 'start_codon':
                info = sep_info(r[8])
                tx_id = [i[1] for i in info if i[0] == 'transcript_id'][0]
                new_r = update_id(r, info, start_codon_cnt[tx_id])
                row = '\t'.join(new_r) + '\n'
                start_codon_cnt[tx_id] += 1
            elif r[2] == 'stop_codon':
                info = sep_info(r[8])
                tx_id = [i[1] for i in info if i[0] == 'transcript_id'][0]
                new_r = update_id(r, info, stop_codon_cnt[tx_id])
                row = '\t'.join(new_r) + '\n'
                stop_codon_cnt[tx_id] += 1

            if r[1] == 'Liftoff' and r[2] not in ['gene', 'transcript']:
                info = sep_info(r[8])
                old_id = [i[1] for i in info if i[0] == 'ID'][0]
                tmp = old_id.split(':')
                if len(tmp) == 2:
                    id_num = 0
                    this_id = old_id
                    repl = False
                elif len(tmp) == 3:
                    id_num = int(tmp[-1])
                    this_id = ':'.join(tmp[:-1])
                    repl = True
                else:
                    print(f'check {old_id}')
                if id_num != liftoff_id_cnt[this_id]:
                    new_r = update_id(r, info, liftoff_id_cnt[this_id], replace=repl)
                    row = '\t'.join(new_r) + '\n'
                liftoff_id_cnt[this_id] += 1

            out.write(row.encode())


gff3_file = sys.argv[1]
out_file = sys.argv[2]
fix_id(gff3_file, out_file)

# run: python fix_gff3.py chm13.draft_v2.0.gene_annotation.gff3 chm13.draft_v2.0.gene_annotation.fixed.gff3.gz

For another gff3:

but I'd stick to the gencode one^

def fix_id(gff3, out):
    id_cnt = set()

    with open(gff3, 'r') as f, open(out, 'w') as out:
        for row in f:
            if row.startswith('#'): 
                out.write(row.encode())
                continue

            r = row.rstrip().split('\t')
            chrom = r[0]
            info = r[8]
            idx, sid = [(i, j) for i,j in enumerate(info) if j[0] == 'ID'][0]
            if sid[1] in id_cnt:
                new_id = f'{chrom}_{sid[1]}'
                info[idx] = (sid[0], new_id)
                r[8] = comb_info(info)
                row = '\t'.join(r) + '\n'
            else:
                id_cnt.add(sid[1])_

            out.write(row.encode())

gff3_file = sys.argv[1]
out_file = sys.argv[2]
fix_id(gff3_file, out_file)

# run: python fix_gff3.py chm13v2.0_RefSeq_Liftoff_v4.gff3 chm13v2.0_RefSeq_Liftoff_v4.fixed.gff3