[mastal]

In trimmomatic's default mode, when it trims adapters from paired end reads, it drops the second read of the pair, because the two reads are reverse complements of each other, so the second read doesn't add any extra information.

In newer versions of trimmomatic this can be turned off, so that it keeps both reads after trimming adapters from paired reads. You need to specify 'TRUE' for the <keepBothReads> parameter of the ILLUMINACLIP command.

ILLUMINACLIP:<fastaWithAdaptersEtc>:<seed mismatches>:<palindrome clip threshold>:<simple clip threshold>:<minAdapterLength>:<keepBothReads>

[tsangkl]

Hi, I found trimmomatic very useful.
And it works well with my Hiseq data using Nextera PE adaptor in single end mode.
But I found the output is quite strange in paired end mode:

My output after trimming:
Input Read Pairs: 12484647 Both Surviving: 4943420 (39.60%) Forward Only Surviving: 7297375 (58.45%) Reverse Only Surviving: 16245 (0.13%) Dropped: 227607 (1.82%)

It seems that the forward and reverse reads after trimming is very unbalanced.
What would cause this?
Thanks.

[Corydoras]

Just in case anybody ever has a similar problem or is confused and stumbles across my post:

Looking closer at my files and where the reverse adapter contamination occurred, it became obvious that the rc sequences were actually simply adapter read through and everything that followed was nonesense which Trimmomatic then perfectly removed. This means in roughly 3% of cases, my fragments were too short for the 150bp HiSeq and the RAD size selection did not work perfectly, but considering it is only 3% and it was my first set of libraries I am fairly happy with that.

Above I stated that I was concerned the forward read would not match the reverse read. Now I believe this is only the case in 100-1000 fragments that consist of tiny fragments with adapter ligating to other tiny fragments of adapter. The majority of the contamination however presents itself in reverse complementary form.

This all obviously rests upon the understanding that when adapter read-through occurs, it will be reverse complementary of the P2 adapters in the forward reads, and reverse complementary of the P1 adapters in the reverse reads. Please feel free to point out if there is something wrong with my logic!

[Brian Bushnell]

I have a comparison of adapter trimming here:

http://seqanswers.com/forums/showthread.php?t=42776

...though it's not published or peer-reviewed. I'll be doing another comparison of quality-trimming soon.

Also, here's a paper comparing quality-trimming methods:

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0085024

[patouch74]

Hi,

I've chosen trimmomatic for my studies about reads processing.
However I have to justify why I choose this tool instead of others pre processing tools.

Do you have any articles which compare trimmomatic to other tools (except trimmomatic's authors article) from what I can get some informations ?

thanks

[Corydoras]

Hi,

Sorry to hijack this post! I got paired-end 150bp RAD sequencing data that I am currently cleaning with Trimmomatic. I just have two quick questions to make sure I am not going wrong anywhere.

1) By the looks of it, most of my adapter contamination occurs within the read. I.e. I have 'P1 -sequence-P1-sequence'. In this case, the single alignment adapter mode will trim this sequence up until the start of the second P1 adapter occurring in the read, leaving me with just 'P1-sequence', am I correct? I am just thinking that I would actually prefer for Trimmomatic to discard these reads entirely, as the reverse part of this read will in all likelihood not come from the same locus as the surviving bit of the forward read? Is there an option to do this? The palindrom mode does not appear to pick these within-read sequences up.

2) Just as a very general question, I also appear to have quite a bit of reverse-complement adapter contamination in my data set and I was wondering if anybody has experience with this for RAD data? Is this something I need to worry about?

Many thanks in advance for any feedback

Sarah

[Brian Bushnell]

The command looks ok, except that you are not providing min adapater length (which defaults to 8). In what way is the output "mysteriously bad"?

Most of the adapters didn't get removed.

[GenoMax]

Hi all,

I'm testing Trimmomatic's performance on adapter removal, and the results are mysteriously bad. So I'd like to make sure I'm not doing anything wrong. This is my command line (modified from the website):

java -Xmx8g -jar trimmomatic-0.32.jar SE -phred33 dirty.fq tclean.fq ILLUMINACLIP:gruseq.fa:2:30:10

...where dirty.fq is a file containing reads with adapter sequences and gruseq.fa is a file containing the adapter sequences. The adapters are inserted synthetically and the reads are tagged, so I know precisely what the correct results should be, and what I'm getting is not really close. Any suggestions?

The command looks ok, except that you are not providing min adapater length (which defaults to 8). In what way is the output "mysteriously bad"?

[Lays Cruz]

How about just running:

#java -jar /usr/local/bin/trimmomatic-0.30.jar PE -threads 4 -phred33 -basein jatoba/Hst_S2_L001_R1_001.fastq -baseout ./jatoba/Hst_S2_trim LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 HEADCROP:18 MINLEN:20

BTW: Are there 3 pairs of samples or are you just providing names of files for holding the output?

No. It comes from a only sample. The other files are the outputs, two PE files with the paired sequences and two SR files with single reads removed from R1 and R2 files.

Much appreciate your suggestion, I will try now.

Thanks.

[GenoMax]

Yes. My data is from Illumina MiSeq and my command's line is as follows:
#java -jar /usr/local/bin/trimmomatic-0.30.jar PE -threads 4 -phred33 ./jatoba/Hst_S2_L001_R1_001.fastq ./jatoba/Hst_S2_L001_R2_001.fastq ./jatoba/fq/Hst_S2_PE_1p.fq ./jatoba/fq/Hst_S2_SR_1p.fq ./jatoba/fq/Hst_S2_PE_2p.fq ./jatoba/fq/Hst_S2_SR_2p.fq LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 HEADCROP:18 MINLEN:20

Thanks.

How about just running:

#java -jar /usr/local/bin/trimmomatic-0.30.jar PE -threads 4 -phred33 -basein jatoba/Hst_S2_L001_R1_001.fastq -baseout ./jatoba/Hst_S2_trim LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 HEADCROP:18 MINLEN:20

BTW: Are there 3 pairs of samples or are you just providing names of files for holding the output?

[Brian Bushnell]

Hi all,

I'm testing Trimmomatic's performance on adapter removal, and the results are mysteriously bad. So I'd like to make sure I'm not doing anything wrong. This is my command line (modified from the website):

java -Xmx8g -jar trimmomatic-0.32.jar SE -phred33 dirty.fq tclean.fq ILLUMINACLIP:gruseq.fa:2:30:10

...where dirty.fq is a file containing reads with adapter sequences and gruseq.fa is a file containing the adapter sequences. The adapters are inserted synthetically and the reads are tagged, so I know precisely what the correct results should be, and what I'm getting is not really close. Any suggestions?

[Lays Cruz]

Are you trimming the files together using the PE option?

Yes. My data is from Illumina MiSeq and my command's line is as follows:
#java -jar /usr/local/bin/trimmomatic-0.30.jar PE -threads 4 -phred33 ./jatoba/Hst_S2_L001_R1_001.fastq ./jatoba/Hst_S2_L001_R2_001.fastq ./jatoba/fq/Hst_S2_PE_1p.fq ./jatoba/fq/Hst_S2_SR_1p.fq ./jatoba/fq/Hst_S2_PE_2p.fq ./jatoba/fq/Hst_S2_SR_2p.fq LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 HEADCROP:18 MINLEN:20

Thanks.

[Lays Cruz]

Yes. My data is Illumina MiSeq and my command line is as follows:
#java -jar /usr/local/bin/trimmomatic-0.30.jar PE -threads 4 -phred33 ./jatoba/Hst_S2_L001_R1_001.fastq ./jatoba/Hst_S2_L001_R2_001.fastq ./jatoba/fq/Hst_S2_PE_1p.fq ./jatoba/fq/Hst_S2_SR_1p.fq ./jatoba/fq/Hst_S2_PE_2p.fq ./jatoba/fq/Hst_S2_SR_2p.fq LEADING:30 TRAILING:30 SLIDINGWINDOW:4:30 HEADCROP:18 MINLEN:20

Thanks for answers.

[GenoMax]

Hi all.
I'm having problems with the trimmomatc outputs, has different numbers of reads in PE files, this should not happen.
Does anyone know what I can do?
Thanks.

Are you trimming the files together using the PE option?

[Lays Cruz]

Hi all.
I'm having problems with the trimmomatc outputs, has different numbers of reads in PE files, this should not happen.
Does anyone know what I can do?
Thanks.