vcfio 1.1.4

A simple and efficient VCF manipulation package.

🔭 Overview

vcfio is an efficient and easy-to-use package for VCF reading, writing and manipluation.
It is designed for robust variant processing, and requires minimum computing resources.

• ⚡ Fast - iterate >4500 variants per second
• 🪶 Lightweight - automatically parses only crucial information about variants (CHROM, POS, ID, REF, ALT, QUAL, FILTER), all the other information (INFO and SAMPLES) is parsed on demand
• 🏁 Efficient - No advanced parsing of lines and casting to hugh memory objects
• 🔌 Dependency free - We do have optional dependencies which enhance the package

🎯 Why ?

The existing python VCF solutions were extremely inefficient - parsing every variant line in advance and casting every bit of data into their own objects or a list of strings, which is very memory-consuming.

This affected our runtime by a huge factor.

We wrote vcfio to overcome those issues, making it a lightweight and dependency-free package.

⚙️ ️Installation

# Basic installation
pip install vcfio

# Include optional dependencies
pip install vcfio[bio]

⭐ Features

:heavy_check_mark: Read and write plain and compressed vcf (no specification needed) - file.vcf, file.vcf.gz, file.vcf.bgz.
:heavy_check_mark: Automatically infer the type of values in the file. For example '1,2,3' will yield [1, 2, 3].
:heavy_check_mark: Parse values on demand for maximum efficiency.
:heavy_check_mark: Fetch variant ranges within a chromosome.

❓ How to Use

VcfReader

Here are some examples of what you can do with VcfReader (We recommend you explore all the available methods)

Click to view usage!

import vcfio

with vcfio.VcfReader('/path/to/file.vcf') as reader:
    # Iterate variants
    for variant in reader:  # type: vcfio.Variant
        print(variant.to_vcf_line())
        
        # Iterate the variant's samples
        for sample_name, sample in variant.samples.items():  # type: AnyStr, EasyDict
            # Try to find "AD" in sample, if not found - find in variant's info, if the value is a dot - return None
            ad = variant.get_value('AD', sample_name, empty_value='.')        
            zygosity = variant.get_zygosity(sample_name)

    # Fetch variants within the range chr3:1-1000
    for variant in reader.fetch('chr3', start=1, end=1000):  # type: vcfio.Variant
        print(variant)

VcfWriter

Here is an example of what you can do with VcfWriter

Click to view usage!

import vcfio

# This will open an existing vcf, introduce a new value to each variant's INFO and write to another vcf.gz
with vcfio.VcfReader('/path/to/output_file.vcf.gz') as reader, \
        vcfio.VcfWriter('/path/to/output.vcf', headers=reader.headers) as writer:  
    for variant in reader:  # type: vcfio.Variant
        variant.info['new_info_field'] = 'new_info_value'
        writer.write_variant(variant)

Variant

Here are some examples of what you can do with Variant (We recommend you to explore all the available methods)

Click to view usage!

import vcfio

variant_line = "chr1	726	.	G	C,T	500	.	DP=200;MQ=250.00	GT:AD:AF:DP:GQ	0/1:10,160,30:0.8,0.15:200:420"

# This will parse the raw variant line into a Variant instance
variant = vcfio.Variant.from_variant_line(variant_line, sample_names=['proband'])
print(
    variant.quality,                        # --> 726
    variant.chromosome,                     # --> "chr1"
    variant.alt,                            # --> ["C", "T"]
    variant.get_zygosity('proband'),        # --> "HET"
    variant.get_value('MQM', 'proband'),    # --> 250
    variant.samples['proband'].get('GT')    # --> "0/1"
)

EasyDict

A Dict-inherited class with a smart get method. It's main (and only) feature is to automatically infer the type of the value it returns. A vcf file rarely specifies the type of values within the variant's data so this object makes a bioinformaticaion's life easier and exempts him from casting-duty. It is used in vcfio.Variant's attributes.

d = EasyDict({
	'simple_int': 1,
	'not_simple_int': '2',
	'this_is_not_a_list': ['abc'],
	'list_of_numbers': ['1', 2, '3.1'],
	'this_is_a_real_list': ['a', 'b', 'c'],
	'dot_is_not_a_value': '.',
    'this_is_a_list': '1,2,3',
    'this_is_a_list_but_i_like_strings': '1,2,3'
}) 
d.get('simple_int')                                                 # --> 1
d.get('not_simple_int', infer_type=True)                            # --> 2
d.get('this_is_not_a_list', infer_type=True)                        # --> 'abc'
d.get('list_of_numbers', infer_type=True)                           # --> [1, 2, 3.1]
d.get('this_is_a_real_list', infer_type=True)                       # --> ['a', 'b', 'c']
d.get('dot_is_not_a_value', empty_value='.', infer_type=True)       # --> None
d.get('this_is_a_list', empty_value='.', infer_type=True)           # --> [1, 2, 3]
d.get('this_is_a_list_but_i_like_strings', empty_value='.')         # --> '1,2,3' 

Credits

This package uses the following open source packages:

• PyVCF

Contributors