genome-fingerprints

Software for creating and comparing genome fingerprints.
More information and datasets: http://db.systemsbiology.net/gestalt/genome_fingerprints/
If you find Genome Fingerprints useful for your work, please cite:
Glusman G, Mauldin DE, Hood L and Robinson M. Ultrafast comparison of personal genomes via precomputed genome fingerprints. Front. Genet. 2017 8:136.

Installing

To install this software, simply clone this repository:
git clone https://github.com/gglusman/genome-fingerprints

This software works best on whole-genome sequencing (WGS) data, in VCF and some other formats. The data type analyzed by this software is variants relative to a reference genome. Therefore, FASTQ, BAM, CRAM and the like are not valid input formats.

Computing and comparing fingerprints

Here are example commands on how to compute fingerprints from genomes that are represented separately, each one in its own VCF, and then how to compare the resulting fingerprints. Anything in bold, please replace with parameter values and file names that make sense for your project.

To create a fingerprint for a genome:
bin/computeDMF.pl myGenome path-to-my-vcfs/myGenome.vcf.gz
...will generate myGenome.outn.gz (and some other files)
To compare two fingerprints:
bin/compareDMFs.pl myFirstGenome.outn.gz mySecondGenome.outn.gz
To serialize fingerprints into a database, using L=120:
bin/serializeDMFs.pl myFingerprintCollection 120 @myListOfFingerprints
bin/serializeDMFs.pl myFingerprintCollection 120 *.outn.gz
To compare a fingerprint to a database:
bin/searchDMFs.pl myGenome.outn.gz myFingerprintCollection
...see the data directory for an example database (CEPH1463 pedigree)
To compare two databases:
bin/searchDMFs.pl aFingerprintCollection anotherFingerprintCollection > setAvsB.aaa
To perform all-against-all comparisons in one database:
bin/searchDMFs.pl aFingerprintCollection > setA.aaa
To find surprises in the comparison:
cat comparison.aaa | bin/findSurprises.pl > surprises.txt

Working with large datasets

To compute fingerprints for a large dataset, like the Thousand Genomes Project (TGP), that is made available as a collection of per-chromosome multi-sample VCFs, use the following protocol. In the commands below, TGP represents the TGP version, TGPdata is the directory where you have your copy of the TGP data, and TGPfp represents the directory where the fingerprints will be stored.

Compute raw fingerprints per chromosome, using L=120 and L=200:
bin/computeDMF1000genomes.pl TGPdata 120,200 TGPfp
Combine the raw fingerprints:
bin/combineDMF1000genomes.pl TGPfp
Normalize them:
bin/normalizeDMFsInDir.pl TGPfp/autosomal
Serialize the normalized fingerprints, using L=200:
bin/serializeDMFs.pl TGPfp/TGP-200 200 TGPfp/autosomal.norm/*.gz
Do all-against-all comparison within the set:
bin/searchDMFs.pl TGPfp/TGP-200 | sort -k3rn | gzip -c > TGPfp/TGP.aaa.gz

For example, if you have the GRCh37 version of the TGP in a directory called TGP37, and you wanted to produce and compare fingerprints with L=200, the actual commands would be:
bin/computeDMF1000genomes.pl TGP37 200 TGP37fp
bin/combineDMF1000genomes.pl TGP37fp
bin/normalizeDMFsInDir.pl TGP37fp/autosomal
bin/serializeDMFs.pl TGP37fp/TGP37-200 200 TGP37fp/autosomal.norm/*.gz
bin/searchDMFs.pl TGP37fp/TGP37-200 | sort -k3rn | gzip -c > TGP37fp/TGP37.aaa.gz
zcat TGP37fp/TGP37.aaa.gz | bin/findSurprises.pl

Note that bin/computeDMF1000genomes.pl can be run in parallel copies to process more than one chromosome at a time, if your system has enough memory. Similarly, bin/normalizeDMFsInDir.pl can be run in concurrent jobs, if needed.

The TGP datasets can be downloaded as follows. Beware, these are big downloads!

gglusman / genome-fingerprints

genome-fingerprints

Installing

Computing and comparing fingerprints

Working with large datasets

About

Languages