Ultrafast Comparison of Personal Genomes via Precomputed Genome Fingerprints.

ABSTRACT: We present an ultrafast method for comparing personal genomes. We transform the standard genome representation (lists of variants relative to a reference) into "genome fingerprints" via locality sensitive hashing. The resulting genome fingerprints can be meaningfully compared even when the input data were obtained using different sequencing technologies, processed using different pipelines, represented in different data formats and relative to different reference versions. Furthermore, genome fingerprints are robust to up to 30% missing data. Because of their reduced size, computation on the genome fingerprints is fast and requires little memory. For example, we could compute all-against-all pairwise comparisons among the 2504 genomes in the 1000 Genomes data set in 67 s at high quality (2

SUBMITTER: Glusman G 

PROVIDER: S-EPMC5623000 | biostudies-literature | 2017

REPOSITORIES: biostudies-literature