Loosing 30-35% of reads after DIAMOND annotation

bob nichols

unread,

Dec 14, 2023, 2:22:06 PM12/14/23

to SAMSA bioinformatics group

Hello,

I was doing a deep dive into the stats from a run as we are comparing 16S depletion kits. I was just curious if people have also seen read losses of around 30-35% after Diamond annotation (read loss from step 3 to step 4). Is this loss assumed to be mRNA that maps to host? I am confused because some reads map to NO HIERARCHY which I assumed would encompass all the unassigned reads.

Thanks!

Bob

Sam Westreich

unread,

Dec 18, 2023, 10:54:45 AM12/18/23

to SAMSA bioinformatics group, bob nichols

Hi Bob,

One area to examine might be checking the read losses in alignments to RefSeq, versus those in SEED Subsystems. Subsystems provides the hierarchy, but it is independent of matches in RefSeq, which is more complete; a read might have no hierarchy match in Subsystems but may have a match in RefSeq, which could help with source identification.

You can also use some of the Python scripts to focus just on reads that have a “NO HIERARCHY” match in Subsystems, finding what they aligned to in RefSeq.

Best,
Sam

--
You received this message because you are subscribed to the Google Groups "SAMSA bioinformatics group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to samsa-bioinformatic...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/samsa-bioinformatics-group/33dcb970-f2bf-4d9f-b687-4f83d3b956edn%40googlegroups.com.

bob nichols

unread,

Dec 18, 2023, 11:29:21 AM12/18/23

to SAMSA bioinformatics group

Hey Sam,

Thank you for the reply! So I did notice that there was some read loss between the refseq and SEED annotated files. I was more curious about the read loss from step 3 to step 4. For example, one of my samples has 21,789,241 reads after ribodepletion but then drops to 15,604,865 reads in the Refseq annotated file (and 10,248,809 SEED subsystem annotated file).

Thanks!

Bob

Sam Westreich

unread,

Jan 12, 2024, 4:09:19 PM1/12/24

to bob nichols, SAMSA bioinformatics group

Hi Bob,

Ah, I see. You may want to look into re-running just the DIAMOND step. DIAMOND has a number of additional options and flags (https://github.com/bbuchfink/diamond/wiki/3.-Command-line-options ) that could be useful, such as the --unal 1 option which outputs unaligned reads from the input file. This could help identify reads that don't get aligned to anything in the DIAMOND RefSeq search.

You could then spot check these through web-based BLAST or another alignment tool of your choice. Or cluster them and see how much uniqueness is in the unaligned fraction.

Best,

Sam

To view this discussion on the web visit https://groups.google.com/d/msgid/samsa-bioinformatics-group/12eefebd-93cb-46ac-8b40-f221a6a3efedn%40googlegroups.com.

--

Sam Westreich, PMP, PhD

Microbiome Scientist, DNAnexus

https://www.dnanexus.com

Incredible Budget

unread,

Jan 8, 2025, 12:24:17 AMJan 8

to SAMSA bioinformatics group

Reply all

Reply to author

Forward