How to exclude chromsomes/small contigs

[mirax87]

Hi Mahmoud,

currently, I am testing JAMM on GRO-seq (SE protocol, Pol2 ChIP-like pattern) in dm6. This genome version contains a lot of small contigs, which I'd like to ignore/exclude from being analyzed. Thus, muy question, how can I tell JAMM to ignore chromosomes or which data do I need to filter (BED, genome_file)?

Cheers,
-Michael

PS: Sorry, if I missed in the documentation

[mahmoudibrahim]

Hi Michael

JAMM considers the chromosomes present in the chromosome size file (-g) only. All other reads are ignored. So you can simply keep in this file the chromosomes you want to analyze only and JAMM will ignore the rest.

Interesting you are using this to call transcripts for GRO-Seq, I would be curious to know what parameters you end up using if you're willing to share this info. You would need to analyze each strand separately because JAMM doesn't call peaks in a stranded manner.

best wishes
Mahmoud

[mirax87]

Hi Mahmoud,

it was about time I came back to you. Time flies...

So, thanks first for the hints - subsetting the -g really speed up the process to a reasonable time (my test runs were too long ago and I don't recall minutes or few hours (< 3 h). ) I agree with the strand-specific peak calling - it's critical, especially in gene-dense organisms like D.melanogaster.

In my tests, I compared MACS2 and JAMM peak calling using 'region' for JAMM and narrowPeak for MACS2 callpeak --nomodel. In the end MACS2 worked slightly better for my purpose (peaks near TSS), for that it was more sensitive.

Overall, it was a close call, but I was under the impression, that both methods are in trouble, when it comes to identifying genes, which are highly transcribed, i.e. show high coverage levels along the entire gene body. Yet, I did not explore all the parameters, I guess the MACS2 --broad could address that. Not sure about JAMM, as I used the JAMM -r region already.

I'll keep looking out for further applications of JAMM in future.

Thanks, sorry for taking so much time, and all the best,
-Michael