Trimming illumina 1.8 reads

[ssharma]

Hi All,
Recently i got data from illumina 1.8.
I am trying to trim bad quality reads, i am using FASTX tool kit for this purpose.
I looked at the quality plot (FASTQC), it shows there are bad quality reads (less than 20) at the end but when i tried to trim it using fastx , what i am getting is untrimmed file, with same read length.
Am i missing something?
This is the command i uded to trim the reads.

fastq_quality_filter -q 20 -p 80 -Q33 -i input.fastq -o output.fastq

Thanks
ssharma

[kmcarr]

Hi All,
Recently i got data from illumina 1.8.
I am trying to trim bad quality reads,....

This is the command i uded to trim the reads.

fastq_quality_filter -q 20 -p 80 -Q33 -i input.fastq -o output.fastq

ssharma,

The program you are using, fastq_quality_filter, does not trim reads, it filters (discards) reads which don't meet the minimums set by your options. The reads which are not filtered will still be their original length. If you want to trim the poor quality bases from the 3' end you need to use fastq_quality_trimmer.

[lshen]

For trimming only, I used
awk ' { if ( length($1)<3 || $1~/^\@HWI/ ) print $0 ; else print substr($0, 1, 85) ;} ' in.fastq > out.fastq

[ssharma]

Thanks Kmcarr,
My bad, i don't know how i missed that.

@Ishen: i will try using this too. Thanks

[kga1978]

You could also try 'prinseq' - I like the feature set in that program for carefully controlling quality

[ssharma]

Hey Kga1978,
Thanks for the suggestion. I just tried prinseq and its really good.
Its doing what i wanted.
But i just have a confusion:
I used input file containing 4 million reads a and used the following command:

perl ~/scripts/FASTAQ/prinseq-lite.pl -fastq in.fastq -trim_qual_left 20 -trim_qual_right 20 -out_format 1 -out_good out_trim

But i am getting only 3.6 million reads with avg length of 125.9

So why i am loosing some reads?

I would really appreciate your help.

Thanks
sshama

[kga1978]

Hi Ssharma,
Trimming down to 20 may be a little too harsh - and you may be losing reads that are overall low quality (which is fine). I usually trim to a quality of ~10-15 and also include some 'average quality' cutoffs. This is my 'standard' setting:
-min_len 20 -ns_max_n 4 -min_gc 10 -max_gc 90 -min_qual_mean 18 -trim_qual_left 10 -trim_qual_right 10

I often lose ~15% on fairly good quality reads.

[drdna]

Don't use FASTX for sequence trimming

FASTX_trimmer is a terrible tool for sequence trimming because it relies on user-defined trimming parameters which are completely arbitrary. Use a tool like LUCY2 or ngs_backbone which trim based on quality parameters.