GenomeKit
What is GenomeKit?
GenomeKit is Deep Genomics’ Python library for fast and easy access to genomic resources such as sequence, data tracks, and annotations. The goal is to let machine learning researchers build data sets easily, and to be creative about how those data sets are designed.
GenomeKit is also designed to work with genome variants, giving users a powerful way to extract features for different genotypes.
In the future it will include other useful resources like secondary structure, conservation, and more.
Useful Features
Fast querying of DNA sequence and genomic tracks.
Fast querying and structured access for annotations (GENCODE, RefSeq, …).
Fast querying and structured access for short read alignments (SAM/BAM files).
Fast querying and filtering for variants (VCF files).
Interval and Variant objects that are convenient, lightweight, and standardized.
Scan the genome for motifs.
Resources Available
Available reference genomes and annotations depend on the data made available by data_manager.
You can get a list of all available genomes and annotations by calling gk_data.data_manager.list_available_genomes().
Some commonly used ones are listed below.
Reference genomes:
hg19 (unpatched GRCh37)
hg19.p13.plusMT (patch 13 with rCRS MT)
hg38 (unpatched GRCh38)
hg38.p12
hg38.p13
hg38.p14
mm10.p6 (GRCm38)
mm39 (unpatched GRCm39)
rn6 (unpatched RGSC 6.0)
macFas5 (unpatched Macaca_fascicularis_5.0)
susScr11 (unpatched SGSC Sscrofa11.1)
Annotations:
GENCODE v19, v29, v41, vM19, vM30, vM31 (basic; comprehensive)
UCSC RefSeq (snapshots)
NCBI RefSeq v105.20190906, v109, v110, m39.v109, Macfas_5.0.v101
ENSEMBL Rnor_6.0.88, Macfas_5.0.95, Sscrofa11.1.98
Next steps
Read the Quickstart tutorial to get started with GenomeKit and study the API Documentation for details. GenomeKit also uses a convention called anchors to disambiguate the alignment of reference and variant sequences that is explained in detail in Anchors.
Contents: