Package — I/O

Reading and writing auxiliary files (LDSC logs, pickles, tabular formats).

`read_ldsc`

Read LDSC output files and parse heritability or genetic correlation results.

Parameters:

Name	Type	Description	Default
`filelist`	`list of str`	List of paths to LDSC output files	`None`
`mode`	`str`	Mode to parse: "h2" for heritability, "rg" for genetic correlation	`"h2"`

Returns:

Type	Description
`DataFrame`	Parsed LDSC results

`load_pickle`

Load a previously saved GWASLab object from a pickle file.

Automatically detects and loads Sumstats, SumstatsPair, or SumstatsMulti objects.

Parameters:

Name	Type	Description	Default
`path`	`str`	File path to the pickle file. Supports `~` for home directory expansion.	required

Returns:

Type	Description
`Optional[Union[Sumstats, SumstatsPair, SumstatsMulti]]`	The loaded GWASLab object (type is automatically detected). Returns None if the file does not exist.

`dump_pickle`

`load_gsf`

Load GWAS sumstats from GSF (GWASLab Standard Format) file.

Parameters:

Name	Type	Description	Default
`path`	`str or Path`	Path to .gsf file or directory (for partitioned)	required
`columns`	`list of str`	Columns to read (None = all columns)	`None`
`filters`	`str`	Filter string for predicate pushdown. Supports: - Operators: ==, !=, <, <=, >, >=, in - Logical operators: & (AND), \| (OR) Examples "P < 5e-8" "CHR == 1" "P < 5e-8 & CHR == 1" "CHR in [1, 2, 3]" "P < 5e-8 \| P > 0.99"	`None`
`verbose`	`bool`	Print progress messages	`True`

Returns:

Type	Description
`Sumstats`	GWASLab Sumstats object

Examples:

>>> mysumstats = gl.load_gsf("sumstats.gsf")
>>> mysumstats = gl.load_gsf("sumstats.gsf", columns=["CHR", "POS", "BETA", "P"])
>>> mysumstats = gl.load_gsf("sumstats.gsf", filters="CHR == 1")
>>> mysumstats = gl.load_gsf("sumstats.gsf", filters="P < 5e-8 & CHR == 1")
>>> mysumstats = gl.load_gsf("sumstats.gsf", filters="CHR in [1, 2, 3]")

`read_gtf`

Fast GTF file reader using Polars.

Returns a pandas DataFrame for compatibility.

Parameters:

Name	Type	Description	Default
`filepath_or_buffer`	`str or buffer`	Path to GTF file (may be gzip compressed) or buffer object	required
`usecols`	`list of str or None`	Restrict which columns are loaded. If None, load all columns. Common columns: seqname, start, end, strand, feature, gene_biotype, gene_id, gene_name	`None`
`features`	`set of str or None`	Drop rows which aren't one of the features in the supplied set (e.g., {'gene', 'transcript', 'exon'})	`None`
`chrom`	`str or None`	Filter by chromosome/seqname early for speed. If None, load all chromosomes. Can be chromosome number (e.g., "1", "23" for X) or name (e.g., "X", "chr1"). For X chromosome, can use "X", "chrX", or "23". For Y chromosome, can use "Y", "chrY", or "24". For MT chromosome, can use "MT", "chrMT", "M", "chrM", or "25".	`None`
`expand_attribute_column`	`bool`	Expand the 'attribute' column into separate columns (default: True)	`True`
`infer_biotype_column`	`bool`	Infer biotype from 'source' column if gene_biotype/transcript_biotype missing	`False`

Returns:

Type	Description
`DataFrame`	DataFrame containing parsed GTF data

`read_bed`

Read a BED file into a pandas DataFrame.

BED files use 0-based, half-open intervals [chromStart, chromEnd).
The chromEnd position is exclusive (not included in the feature).

Parameters:

Name	Type	Description	Default
`bed_path`	`str`	Path to BED file. Supports uncompressed (.bed) and gzipped (.bed.gz) files.	required
`usecols`	`list of int`	Column indices to read (0-based). If None, reads all columns. Default columns: chrom (0), chromStart (1), chromEnd (2), and optional fields.	`None`
`header`	`bool`	Whether the file has a header line. BED files typically don't have headers, but custom tracks may have "browser" or "track" lines. If None, auto-detects by checking for "browser" or "track" lines.	`None`
`comment`	`str`	Character(s) that indicate comment lines to skip. Default is "#". BED custom tracks may have "browser" or "track" header lines.	`None`
`verbose`	`bool`	If True, prints information about the file being read.	`False`
`log`	`Log`	Logger instance for messages.	`None`

Returns:

Type	Description
`DataFrame`	DataFrame with BED data. Column names depend on number of columns: - BED3: chrom, chromStart, chromEnd - BED4+: chrom, chromStart, chromEnd, name, ... Columns are named according to BED specification.

Notes

BED format specification:
- chrom: Chromosome name (e.g., chr1, chrX, 1, X)
- chromStart: Starting position (0-based)
- chromEnd: Ending position (1-based, exclusive)
- Optional fields follow the standard BED12 format

Coordinate system:
- BED uses 0-based, half-open intervals [chromStart, chromEnd)
- For example, chromStart=0, chromEnd=100 spans bases 0-99
- In 1-based coordinates, this corresponds to positions 1-100

References

UCSC BED format: https://genome.ucsc.edu/FAQ/FAQformat.html#format1