Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Animal Genomics and Improvement Laboratory » Research » Publications at this Location » Publication #323634

Title: Fast single-pass alignment and variant calling using sequencing data

item Vanraden, Paul
item Bickhart, Derek

Submitted to: Plant and Animal Genome Conference Proceedings
Publication Type: Abstract Only
Publication Acceptance Date: 1/8/2016
Publication Date: 1/8/2016
Citation: Van Raden, P.M., Bickhart, D.M. 2016. Fast single-pass alignment and variant calling using sequencing data. Plant and Animal Genome Conference Proceedings. San Diego, CA, Jan. 9–13, W161.

Interpretive Summary:

Technical Abstract: Sequencing research requires efficient computation. Few programs use already known information about DNA variants when aligning sequence data to the reference map. New program findmap.f90 reads the previous variant list before aligning sequence, calling variant alleles, and summing the allele counts for each DNA source in a single pass. Advantages are faster processing, more precise alignment, more useful data summaries, more compact output, and fewer steps. Programs findmap and BWA were compared using simulated paired end reads of length 150 from fragments of length 1,000 at random locations within the UMD3.1 bovine reference assembly. Each base had 1% probability of error and 1% probability of missing. The 39 million variants from run 5 of the 1,000 bull genomes project were included, with every other variant set to reference or alternate. With 1 processor, BWA required 629 minutes per 1X for alignment, whereas findmap required 2 minutes per 1X for alignment and variant calling. Percentage of correctly mapped reads was 90.5% from BWA and 92.9% from findmap. Variant calls were output by findmap only for the 88.2% of pairs where both ends were located within the fragment length and of opposite orientation. Percentages of variants called correctly were 99.8% for SNPs and 99.9% for deletions, while insertions had 99.9% of alternate calls correct but only 98.6% of reference calls. Memory required by BWA was 4.6 Gbytes / processor, whereas findmap required 46 Gbytes that could be shared by multiple processors. Simultaneous alignment and variant calling is an efficient and accurate strategy.