MG-RAST pipeline 4.0.2 released with new rRNA detection mechanism
We have upgraded our production system to use a new pipeline.
In our testing the results are very similar to the existing approach and the delta is almost entirely computationally, there little to no effects on the results.
We routinely monitor new tools that offer identical or improved sensitivity and specificity while adding run time benefits (e.g. less CPU or RAM). We have found a new way to identify potential rRNA genes: SortMeRNA.
In our testing the speed up was 3-10x depending on the length of the input sequences.
We are now using SortMeRNA for rRNA identification instead of vsearch (version 4.0.1 used "vsearch --strand both --usearch_global <input> --id 0.70 --db <m5rna_reduced>"), version 4.0.2 is now using the command "sortmerna -e 0.1 --blast '1 cigar qcov qstrand' --reads <input> --ref <m5rna_reduced>" to identify potential rRNA genes.
The <m5rna_reduced> 70% identity clustered version of m5rna database.
In our testing the results are very similar to the existing approach and the delta is almost entirely computationally, there little to no effects on the results.
We routinely monitor new tools that offer identical or improved sensitivity and specificity while adding run time benefits (e.g. less CPU or RAM). We have found a new way to identify potential rRNA genes: SortMeRNA.
In our testing the speed up was 3-10x depending on the length of the input sequences.
We are now using SortMeRNA for rRNA identification instead of vsearch (version 4.0.1 used "vsearch --strand both --usearch_global <input> --id 0.70 --db <m5rna_reduced>"), version 4.0.2 is now using the command "sortmerna -e 0.1 --blast '1 cigar qcov qstrand' --reads <input> --ref <m5rna_reduced>" to identify potential rRNA genes.
The <m5rna_reduced> 70% identity clustered version of m5rna database.
Comments
Post a Comment