liftover.py

from pathlib import Path
import sys

from pyliftover import LiftOver

from conflict_positions import Region

LIFTOVER_DIR = Path("/confidential/tGenVar/ref/liftover")
HG38_T2T_v1_0_CHAIN = LIFTOVER_DIR / "hg38_to_t2t/FINAL/merged.over.chain"
T2T_v1_0_TO_v1_1_CHAIN = LIFTOVER_DIR / "t2t_v1.0_to_v1.1/t2t_v1.0_to_v1.1.chain"

# preload the chain files for better caching
lo_1 = LiftOver(str(HG38_T2T_v1_0_CHAIN))
lo_2 = LiftOver(str(T2T_v1_0_TO_v1_1_CHAIN))

def lift(lo, region: Region):
    chrom, start, end = region.vals()
    conversions_1 = lo.convert_coordinate(chrom, start)
    conversions_2 = lo.convert_coordinate(chrom, end)
    if not conversions_1 or not conversions_2:
        return None
    else:
        return Region(chrom, conversions_1[0][1], conversions_2[0][1])

def lift_hg38_to_t2t_v1_0(region: Region):
    return lift(lo_1, region)


def lift_t2t_v1_0_to_v1_1(region: Region):
    return lift(lo_2, region)


def lift_hg38_to_t2t_v1_1(region: Region):
    r = lift_hg38_to_t2t_v1_0(region)
    if r is None:
        return None
    else:
        return lift_t2t_v1_0_to_v1_1(r)

if __name__ == "__main__":
    regions = sys.argv[1:]

    for r in regions:
        print(lift_hg38_to_t2t_v1_1(Region.from_str(r)))