DNA aligner accuracy: BWA, Bowtie, Soap and SubRead tested with simulated reads
In the past few posts, we've looked at RNA-seq aligner performance in terms of accuracy and speed. In this post, I'll take a look at the accuracy of DNA aligners using simulated reads.
The first step is to download the genome of interest. I'm using Arabidopsis as its pretty small and good for quick benchmarking. I downloaded the genome from Ensembl plant ftp site (link).
Next step was to generate simulated reads that uniformly cover the genome at user-selected length and intervals. I couldn't find any previously made simulators to do exactly that, so I chained together some awk and bedtools commands (code at the bottom of the post) to generate the reads. I generated pseudo reads of 50, 100 & 200 bp in length at 10 bp intervals and output them in fasta format. Here is an example.
>1-0-50
CCCTAAACCCTAAACCCTAAACCCTAAACCTCTGAATCCTTAATCCCTAA
>1-10-60
TAAACCCTAAACCCTAAACCTCTGAATCCTTAATCCCTAAATCCCTAAAT
>1-20-70
ACCCTAAACCTCTGAATCCTTAATCCCTAAATCCCTAAATCTTTAAATCC
T…
The first step is to download the genome of interest. I'm using Arabidopsis as its pretty small and good for quick benchmarking. I downloaded the genome from Ensembl plant ftp site (link).
Next step was to generate simulated reads that uniformly cover the genome at user-selected length and intervals. I couldn't find any previously made simulators to do exactly that, so I chained together some awk and bedtools commands (code at the bottom of the post) to generate the reads. I generated pseudo reads of 50, 100 & 200 bp in length at 10 bp intervals and output them in fasta format. Here is an example.
>1-0-50
CCCTAAACCCTAAACCCTAAACCCTAAACCTCTGAATCCTTAATCCCTAA
>1-10-60
TAAACCCTAAACCCTAAACCTCTGAATCCTTAATCCCTAAATCCCTAAAT
>1-20-70
ACCCTAAACCTCTGAATCCTTAATCCCTAAATCCCTAAATCTTTAAATCC
T…