dynast count very slow - may not be parallelized?

[wgao688]

Hi Xiaojie,

Thank you for making the very useful dynast pipeline. We use it a lot in our analysis. One issue I have running into is that dynast count is very slow. I am using dynast v1.01 and for a standard deep sequencing run (NextSeq, Novaseq, etc.) a library will take 4-5 days (96+ hours) to complete. I am wondering if it is somehow related to not using the number of threads specified (I specify -t 8-16). When I check my running processes using ps -u, I sometimes see that there is only 1 process running for dynast count when there were the number t specified for dynast consensus, which runs much faster (<8 hours).

Here is a standard workflow that I use:

dynast align -i $star_index_directory -o $align_directory -x 10xv3 $fastq_seq_file $fastq_umi_file -w $whitelist --STAR-overrides '--soloFeatures Gene GeneFull --limitBAMsortRAM 60000000000'
dynast consensus -t 16 -g $gtf_file --barcode-tag CB --umi-tag UB -o consensus Aligned.sortedByCoord.out.bam
dynast count -t 16 -g $gtf_file --barcode-tag CB --umi-tag UB -o count_with_consensus --barcodes Solo.out/GeneFull/raw/barcodes.tsv --conversion TC consensus/consensus.bam

Thanks for your help!
William