Looking for a trimming software that does these things

[antifolate]

Hello,

I'm looking for a trimming/filtering software that can do the following:

1) Trim both ends until there's at least a certain number of consecutive bases higher than a specific quality score.

2) Remove the 3'-regions of a certain length if they contained a certain percentage of bp below a specific quality score. For example, remove 3' ends of 200 bp if they were made of more than 10% of bp below 20 phred score.

3) Filter out reads with a certain percentage of bp below a specific quality score.

4) Remove reads with a certain number of consecutive Ns.

5) Be paired-end-aware, i.e. if one read was removed, remove its pair (there're several of these available, but without the other features).

6) If a read was identical to the reverse compliment to its pair, remove it.

I'd really appreciate your help.

[GenoMax]

BBduk.sh (part of BBMap), Trimmomatic, Cutadapt (and perhaps others that I am missing) should fit the bill. Though they may not check every box you have up there they should get the job done.

[antifolate]

Thanks. I tried Trimmomatic but not the other two. BBduk.sh seems promising (so does the BBMap package), but I'm gonna have to take a while before understanding its syntax. I'll post back if it does what I want.

[Brian Bushnell]

Hello,

I'm looking for a trimming/filtering software that can do the following:

1) Trim both ends until there's at least a certain number of consecutive bases higher than a specific quality score.

BBDuk used to use this strategy, but it's not optimal so I don't really recommend it. I was able to demonstrate empirically that it was not too good, either. So, BBDuk currently uses the Phred algorithm for quality trimming, which is optimal, though it's technically possible to disable that with a flag and use the old method instead. BBDuk also supports windowed trimming (trim until the average in a sliding window exceeds some threshold).

3) Filter out reads with a certain percentage of bp below a specific quality score.

The "maq" flag filters by average quality, where average quality is calculated by transforming the quality scores into probabilities, so basically if you set "maq=20" it removes reads with an expected error rate greater than 1%. I don't recommend setting it that high, though.

4) Remove reads with a certain number of consecutive Ns.

The "maxns=X" flag will filter reads with at least X Ns, but it doesn't care whether they are consecutive.

5) Be paired-end-aware, i.e. if one read was removed, remove its pair (there're several of these available, but without the other features).

Check.

6) If a read was identical to the reverse compliment to its pair, remove it.

You can do this with BBMerge, by running it but telling it not to join overlapping reads (using the "join=f" flag), and using the "maxlength" flag plus the "out" and "outu" streams. "maxlength=X" will send reads with insert sizes longer than X to outu rather than out. So:

bbmerge.sh in=reads.fq out=short.fq outu=long.fq join=f maxlen=150

(this command assumes pairs are interleaved in one file)

[antifolate]

I just got around to trying these commands and- although they're not exactly what I'm trying to do- they worked pretty well. bbmerge would merge my reads so I avoided it.

Thank you!

[relipmoc]

try skewer

Another option is skewer. Good luck!

I just got around to trying these commands and- although they're not exactly what I'm trying to do- they worked pretty well. bbmerge would merge my reads so I avoided it.

Thank you!

[antifolate]

@Brian

"... though it's technically possible to disable that with a flag and use the old method instead."

How can I do this?

[Brian Bushnell]

@Brian

"... though it's technically possible to disable that with a flag and use the old method instead."

How can I do this?

Add the flag "otm=f" (otm stands for "optimal trimming mode").

[antifolate]

otm=f (outputtrimmedtomatch) Output reads trimmed to shorter
than minlength to outm rather than discarding.

What bbduk you talking about?

[Brian Bushnell]

Ooops, looks like I have an overloaded flag. Thanks for spotting that! I'll rename that one to "ottm" in the next release. Currently, "otm" acts on the quality trimming, so "outputtrimmedtomatch" would have to be fully spelled out in order to function according to that description. To be more specific for now, use the flag "optitrim=f" to turn off optimal trimming, and "outputtrimmedtomatch" to dictate whether trimmed reads shorter than minlen go to outm.

[antifolate]

I didn't know bbduk was your work. Thanks for the help and the tool!