import logging
from typing import Dict, List, Union
# Variation of https://github.com/epiviz/epivizfileserver/src/epivizfileserver/cli.py
__author__ = "jkanche"
__copyright__ = "jkanche"
__license__ = "MIT"
def _parse_all_attribute(row: str) -> Dict:
"""Extract all keys from the gtf/gff attribute string.
Args:
row:
A row from GTF.
Returns:
A dictionary containing extracted keys and their values.
"""
attr = row["group"]
infos = attr.split(";")
vals = {}
for info in infos:
if info == "" or len(info) == 0:
continue
imap = info.strip().split(" ", 1)
vals[imap[0]] = imap[1].strip().strip('"')
return {**row, **vals}
[docs]
def parse_gtf(
path: str,
compressed: bool,
skiprows: Union[int, List[int]] = None,
comment: str = "#",
):
"""Read a GTF file as :py:class:`~pandas.DataFrame`.
Args:
path:
Path to the GTF file.
compressed:
Whether the file is gzip compressed.
skiprows:
Rows to skip if the gtf file has header.
comment:
Character indicating that the line should not be
parsed. Defaults to "#".
Returns:
Pandas DataFrame containing annotations from GTF.
"""
from joblib import Parallel, delayed
from pandas import DataFrame, read_csv
logging.info(f"Reading File - {path}")
if compressed:
df = read_csv(
path,
sep="\t",
names=[
"seqnames",
"source",
"feature",
"starts",
"ends",
"score",
"strand",
"frame",
"group",
],
compression="gzip",
skiprows=skiprows,
comment=comment,
)
else:
df = read_csv(
path,
sep="\t",
names=[
"seqnames",
"source",
"feature",
"starts",
"ends",
"score",
"strand",
"frame",
"group",
],
skiprows=skiprows,
comment=comment,
)
rows = Parallel(n_jobs=-2)(delayed(_parse_all_attribute)(row) for _, row in df.iterrows())
gtf = DataFrame.from_records(rows)
gtf.drop(["group"], axis=1)
return gtf
[docs]
def read_gtf(
file: str,
skiprows: Union[int, List[int]] = None,
comment: str = "#",
) -> "GenomicRanges":
"""Read a GTF file as :py:class:`~genomicranges.GenomicRanges.GenomicRanges`.
Args:
file:
Path to GTF file.
skiprows:
Rows to skip if the gtf file has header.
comment:
Character indicating that the line should not be
parsed. Defaults to "#".
Returns:
Genomic Ranges with annotations from the GTF file.
"""
compressed = True if file.endswith("gz") else False
data = parse_gtf(file, compressed=compressed, skiprows=skiprows, comment=comment)
from ..GenomicRanges import GenomicRanges
return GenomicRanges.from_pandas(data)