Bwa Alignment And Input Pair-End Format, Does Fastq Encoding Of The Input Matter?

Question

[Tonyzeng]

Hi, I am curious about a question. I have pair-end reads here as T1.fq and T2.fq (fastq format) that are produced by illumina 1.8. When I do BWA alignment with galaxy, I need to convert T1 and T2 format to sangerfastq format for no doubt by Galaxy groomer. and I got good mapping result.

My question is that, when I do the same BWA alignment under Linux but not Galaxy, do I need to use T1 and T2 that have already been converted to sangerfastq format?

(When I did quality filtering and trimmer on T1 and T2 reads, I added -Q33 on the end of linux commands and they works good. can I add -Q 33 on BWA aln commands but not need go to galaxy groomer to convert format before I do BWA. you know, it already takes a long time to do galaxy groomer.

$ bwa aln mm10.fa  T1.fq -Q33 > T1.sai

$ bwa aln mm10.fa  T2.fq -Q33 > T2.sai

Answer 1

[Istvan Albert]

You can perform quality encoding shifting with seqtk at extremely high speeds most likely thousands of times faster than the FastQ groomer:

seqtk seq -Q 33 input.fq > output.fq

[Ashutosh Pandey]

Here is the thread that has already discussed this problem.

Convert Illumina reads to Sanger score

Emboss seqret seems to be the best based on votes.

[Tonyzeng]

Thanks Albert, that's so helpful!

[Istvan Albert]

ok wait, just tested this out to make sure - it does not seem to work,

I recalled this feature from memory it is named "quality shift" but does not seem to do anything, maybe I am misusing it

Answer 2

[Ashutosh Pandey]

It will always be good to convert your read quality to phred+33 or sanger fastq format. The "-q" command and not the big "-Q" is used to trim low quality based from the end of the read in bwa. You should feed Sanger format fastq to BWA aligner.

[Tonyzeng]

thank you for the answer, Ashutoshmits