Exogenous Reads Not Being Discarded After Splitting

[alessandro-vai]

Hi!

First, I want to say that I really enjoy using SpikeFlow!

I’ve encountered an issue while working with paired-end data. I noticed that some exogenous reads remain in the reference BAM file after the splitting step. Specifically, this seems to happen in cases where the read1 maps to a reference chromosome, but read2 maps to an exogenous chromosome.

Shouldn't these reads be discarded as well?

Thank you in advance.

Alessandro

[DavideBrex]

Hi Alessandro,

Thank you, I am happy to hear that you find SpikeFlow useful.

Regarding the read split, SpikeFlow does not handle the situation you are pointing out as it is now.

While testing several paired-end samples with Spiker (see here), I noticed that in the quality control table generated (columns: unmapped_reads, qcfail_reads, duplicate_reads, secondary_reads, low_maq, diff_genome), all the values were zeros (or neglectable). diff_genome is the number of discordant mates, meaning those aligning on different genomes. The only exception was low_maq, which had many reads and it could impact the normalization factor calculation. Consequently, I only implemented the mapq filter in SpikeFlow.

Anyway, it might be helpful to keep those metrics in case an experiment goes wrong and you find high levels of discordant mates.

I am working on a new version of SpikeFlow, which will generate the signal tracks for the spike-in for QC purposes (as recently pointed out in this nature commentary. I will also introduce a count and removal of the discordant mates as you suggested, although, as I said, it should not impact that much the final outcomes of the analysis for most ChIP-Rx experiments.

Best,

Davide

[alessandro-vai]

Cool, indeed the number of those reads is limited. It was just to let you know.

Good to hear that you are working of novel features!

All the best, Alessandro

---

From: Davide Bressan @.> Sent: Wednesday, October 2, 2024 11:33:25 AM To: DavideBrex/SpikeFlow @.> Cc: alessandro-vai @.>; Author @.> Subject: Re: [DavideBrex/SpikeFlow] Exogenous Reads Not Being Discarded After Splitting (Issue #22)

Hi Alessandro,

Thank you, I am happy to hear that you find SpikeFlow useful.

Regarding the read split, SpikeFlow does not handle the situation you are pointing out as it is now.

While testing several paired-end samples with Spiker (see herehttps://github.com/liguowang/spiker/blob/main/bin/split_bam.py), I noticed that in the quality control table generated (columns: unmapped_reads, qcfail_reads, duplicate_reads, secondary_reads, low_maq, diff_genome), all the values were zeros (or neglectable). diff_genome is the number of discordant mates, meaning those aligning on different genomes. The only exception was low_maq, which had many reads and it could impact the normalization factor calculation. Consequently, I only implemented the mapq filter in SpikeFlow.

Anyway, it might be helpful to keep those metrics in case an experiment goes wrong and you find high levels of discordant mates.

I am working on a new version of SpikeFlow, which will generate the signal tracks for the spike-in for QC purposes (as recently pointed out in this nature commentaryhttps://www.nature.com/articles/s41587-024-02377-y. I will also introduce a count and removal of the discordant mates as you suggested, although, as I said, it should not impact that much the final outcomes of the analysis for most ChIP-Rx experiments.

Best,

Davide

— Reply to this email directly, view it on GitHubhttps://github.com/DavideBrex/SpikeFlow/issues/22#issuecomment-2388043913, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APKTLVK4XWI36CXWROERGNDZZO4WLAVCNFSM6AAAAABPFBFWYSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOBYGA2DGOJRGM. You are receiving this because you authored the thread.Message ID: @.***>