readpaf is a fast parser for minimap2 PAF (Pairwise mApping Format) files. It is written in pure python with no required dependencies unless a pandas DataFrame is required.
Minimal install:
pip install readpafWith optional pandas dependency:
pip install readpaf[pandas]Direct download
As readpaf is a self contained module it can be installed by downloading just the module. The latest version is available from:https://raw.githubusercontent.com/alexomics/read-paf/main/readpaf.py
or a specific version can be downloaded from a release/tag like so:
https://raw.githubusercontent.com/alexomics/read-paf/v0.0.5/readpaf.pyPyPI is the recommended install method.
readpaf only has one user function, parse_paf that accepts of file-like object; this
is any object in python that has a file-oriented API (sys.stdin, stdout from subprocess,
io.StringIO, open files from gzip or open).
The following script demonstrates how minimap2 output can be piped into readpaf
from readpaf import parse_paf
from sys import stdin
for record in parse_paf(stdin):
    print(record.query_name, record.target_name)readpaf can also generate a pandas DataFrame:
from readpaf import parse_paf
with open("test.paf", "r") as handle:
    df = parse_paf(handle, dataframe=True)readpaf has a single user function
parse_paf(file_like=file_handle, fields=list, na_values=list, na_rep=numeric, dataframe=bool)Parameters:
- file_like: A file like object, such as sys.stdin, a file handle from open or io.StringIO objects
- fields: A list of 13 field names to use for the PAF file, default:
These are based on the PAF specification."query_name", "query_length", "query_start", "query_end", "strand", "target_name", "target_length", "target_start", "target_end", "residue_matches", "alignment_block_length", "mapping_quality", "tags" 
- na_values: A list of values to interpret as NaN. This is only applied to numeric fields, default: ["*"]
- na_rep: Value to use when a NaN value specified in na_valuesis found. This should ideally be0to match minimap2's output default:0
- dataframe: bool, if True, return a pandas.DataFrame with the tags expanded into separate Series
If used as an iterator, then each object returned is a named tuple representing a single line in the PAF file.
Each named tuple has field names as specified by the fields parameter.
The SAM-like tags are converted into their specified types and stored in a dictionary with the tag name as the key and the value a named tuple with fields name, type, and value.
When print or str are called on PAF record (named tuple) a formated PAF string is returned, which is useful for writing records to a file.
The PAF record also has a method blast_identity which calculates the blast identity for that record.
If used to generate a pandas DataFrame, then each row represents a line in the PAF file and the SAM-like tags are expanded into individual series.