Problem with Circle_finder microDNA.RunAsOneJob.parallel.bowtie.sh script

[jiajinlongkang]

Hi Pankaji,

This is Jack. Recently I am using the Circle_finder pipeline to identify eccDNA based on single cell ATAC-seq data. The input I used is the fastq file of reads pooled together from all single cells. I have successfully run the tutorial code on our sample, but when I applied the pipeline to our data, there is such strange problem that the program becomes stuck at some time point. To be more specific, you know that when the pipeline is running, it will create and update a bunch of directories named by a chromosome regions. The problem is that at certain time point these chromosome directories stop to be generated and updated. That is, the program is stuck indefinitely.

After some extensive trouble shooting, I found that the problem lies in this script (microDNA.RunAsOneJob.parallel.bowtie.sh). I wanted to identify ecDNAs which can be as long as 1-3 Mb, so I specified the “Longest circle wish to identify” as 5000000. It turns out I got a few regions longer than 1 Mb in the Island.fa file, and when the while loop in microDNA.RunAsOneJob.parallel.bowtie.sh iterates through these long regions, it meets a “segmentation fault (core dumped)” error. It appears as a system resource overflow error. I fount out that when this error happens the program abolishes this iteration and moves to the next, but with the accumulation of it, the program will eventually be stuck.

It is very puzzling to me why this would happen. We are really interested in identifying long ecDNA over 1Mb but this problem prevents us from doing so. Is there a way for Cicle_finder to get around this issue? Can you take a look at it? I really appreciate your help!

Thanks,
Jack