search

alexdobin / STAR

RNA-seq aligner
MIT License
1.84k stars 504 forks source link

Chimeric reads alignment. Some threads take much longer to complete. #2103

Open alexfriman opened 6 months ago

alexfriman commented 6 months ago

I run STAR 2.7.10b with the following command: STAR --runThreadN 72 --runMode alignReads --genomeDir /home/frimanav/data/genomes/dm6.standardcontigs/ --outSAMtype BAM Unsorted --limitBAMsortRAM 30000000000 --outFilterMultimapNmax 10000 --outFilterMismatchNmax 8 --outReadsUnmapped Fastx --outSAMattributes NH HI NM MD AS --outSAMmultNmax 500 --alignWindowsPerReadNmax 100000 --seedPerWindowNmax 500 --alignTranscriptsPerWindowNmax 1000 --winAnchorMultimapNmax 500 --alignTranscriptsPerReadNmax 500000 --outFileNamePrefix SRR12655953_Aligned_25nt_MM_sbatch --chimSegmentMin 20 --chimMainSegmentMultNmax 500 --chimMultimapNmax 500 --chimFilter None --readFilesIn SRR12655953_1.fastq SRR12655953_2.fastq

CPU usage plot looks like this:

image

A lot of threads are completed after 3 days, but after that it takes much longer for a thread to finish the task, and each next day has less and less completed thread events.

What can be the reason?

alexdobin commented 6 months ago

Hi @alexfriman

You are using very aggressive mapping parameters: --alignWindowsPerReadNmax 100000 --seedPerWindowNmax 500 --alignTranscriptsPerWindowNmax 1000 --winAnchorMultimapNmax 500 --alignTranscriptsPerReadNmax 500000 With these parameters some reads take very long time to map, and so some threads get stuck.