Trimmomatic trims most of the reverse strand

Question

I have a paired-end data. It is from small part of human genome not a WGS or WES. I use BWA for alignment and do variant calling. Since I had some false positives I wanted to do trimming.

I used trimmomatic with command:

trimmomatic PE -phred33 -threads 2 read1.fastq.gz read2.fastq.gz read1_trimmed.fastq read2_trimmed.fastq LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15`

I do not know why but I have this output:

Input Read Pairs: 1834513 
Both Surviving: 1822034 (99.32%) 
Forward Only Surviving: 5740 (0.31%)
Reverse Only Surviving: 417 (0.02%)
Dropped: 6322 (0.34%)
TrimmomaticPE: Completed successfully

I do not have any adapters since the are removed in bcl2fastq step. In FASTQ quality control I have seen that quality of reverse and forward reads are similar in quality but trimmomatic trimmed many reads in reverse file. I cannot do alignment with these files, BWA gives a warning that says:

[bseq_read] the 2nd file has fewer sequences. 
[process] read 155012 sequences (16693324 bp)... 
[bseq_read] the 2nd file has fewer sequences.

Is there anyone who faced a problem like this before?

Thank you in advance

Answer 1

Maybe you need to remove the unpaired reads:

trimmomatic PE -phred33 -threads 2 \
    read1.fastq.gz read2.fastq.gz \
    read1_trimmed.fastq.gz read1_unique.fastq.gz \
    read2_trimmed.fastq.gz read2_unique.fastq.gz \
    LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15

Then you can try with bwa.