Is possible to rescue the reads trimmet out with Trimmomatic?

[g.galazzo]

Hello to everyone, i'm new of this forum so i will apologize if could be a thread for this argument (i searched but i can't find it ).
Here is my question,
using Trimmomatic v0.10.1 is possible to rescue all that reads that have been trimmed out by the filteres( for example AVGQUAL)?
if yes, could you tell me how?
or if it's not possible could you give me some advice about other programs that can perform that kind of task?
thank you all for the help

ciao a tutti!

[Brian Bushnell]

BBDuk will do that. For example, if you want to remove phiX reads, and reads with average quality below 10:

bbduk.sh ref=phix.fasta k=31 maq=10 in=reads.fq out=good.fq outm=bad.fq

Passing reads go to out, and failing reads go to outm. I'm not sure whether you can do this with Trimmomatic; the manual seems to indicate that you can save pairs where one fails and the other doesn't, but does not seem to mention a way to catch all filtered reads.

[westerman]

As Brian implied, Trimmomatic will separate the reads into R1_paired, R2_paired, R1_unpaired (e.g., R1 reads that no longer have an R2 mate) and R2_unpaired. Failed reads are thrown away.

I don't use BBDuk -- too lazy to switch from Trimmomatic -- but I think that Brian's example will only produce two files -- one when *both* pairs pass the given parameters (in this case no-phiX, avg quality >= 10) and another file that contains pairs when *either* of the pairs fails. It is also possible to set up BBDuk to output 4 files ala Trimmomatic.

However while the Trimmomatic and BBDuk "pass" files should be effectively the same (given algorithmic differences) the "fail" files will be different with Trimmomatic only having single reads and BBDuk having paired reads.

This is a long winded post that says, yes, use BBDuk if you really want all of the failed reads with all of their warts.

[Brian Bushnell]

...I think that Brian's example will only produce two files -- one when *both* pairs pass the given parameters (in this case no-phiX, avg quality >= 10) and another file that contains pairs when *either* of the pairs fails.

That's true, though note that there is also a "reib" ("removeifeitherbad") flag which will switch the behavior between removing pairs if either fails, and removing only if both fail.

And it's also true that there's not much point in storing reads that fail quality metrics (unless you want to do some kind of QC analysis on them); it makes more sense when used for kmer filter.