Tweet
Hi All,
I used bowtie to map reads with k = 10 (allow at most 10 alignments for one read). I hoped to get some statistical result for mapping but found the summary of bowtie result was not consistent with samtools flagstats. It seems that samtools flagstats and picard ONLY support sam file with ONE alignment. In fact, they can only work on ALIGNMENT rather than READ.
For example, I got ~25% reads mapped from bowtie log but ~56% from samtools flagstats and picard. Since one read may have multiple alignments, samtools and picard treat each alignment as a mapped read! The confusing thing is the label for the result claims it is calculated based on READ! For example, PCT_PF_READS_ALIGNED is used in picard, but it will mislead you, since its value is the percentage of number of valid alignments NOT reads themselves.
I do think this maybe a bug if the label claims as such?
Any answers or explanations? Thanks.
I used bowtie to map reads with k = 10 (allow at most 10 alignments for one read). I hoped to get some statistical result for mapping but found the summary of bowtie result was not consistent with samtools flagstats. It seems that samtools flagstats and picard ONLY support sam file with ONE alignment. In fact, they can only work on ALIGNMENT rather than READ.
For example, I got ~25% reads mapped from bowtie log but ~56% from samtools flagstats and picard. Since one read may have multiple alignments, samtools and picard treat each alignment as a mapped read! The confusing thing is the label for the result claims it is calculated based on READ! For example, PCT_PF_READS_ALIGNED is used in picard, but it will mislead you, since its value is the percentage of number of valid alignments NOT reads themselves.
I do think this maybe a bug if the label claims as such?
Any answers or explanations? Thanks.