Filtering Illumina GAIIx reads using fastx

[rubi]

Hi,

I'm an NGS newbie.
I'm trying to filter a paired-end fastq reads file produced by Illumina GAIIx, using fastx's default (fastq_quality_filter -q 20).

I would expect this to filter reads such as:

@HWUSI-EAS1676_0037_FC:7:1:1250:1070#0/1
GTATGNGGATATTATTATTTTTTTGTTTAAATTGTGTATAANTNNNTNNNNTAGATGGCTTTCTGTNNNNNNNNN
+HWUSI-EAS1676_0037_FC:7:1:1250:1070#0/1
ZUSSYB]YT\\ba``dafffgggeggggcagfgg]fccfdQBZBBBXBBBBUZXV_aa^^dfWdfBBBBBBBBBB

but it does not.

Anyone knows what am I doing wrong?

[Jenzo]

I would suggest that your quality data is encoded in Illumina V1.3+, I think the fastx toolkit needs sanger encoded quality values..

Galaxy Groomer turned your sample read into:

@HWUSI-EAS1676_0037_FC:7:1:1250:1070#0/1
GTATGNGGATATTATTATTTTTTTGTTTAAATTGTGTATAANTNNNTNNNNTAGATGGCTTTCTGTNNNNNNNNN
+HWUSI-EAS1676_0037_FC:7:1:1250:1070#0/1
;644:#>:5==CBAAEBGGGHHHFHHHHDBHGHH>GDDGE2#;###9####6;97@BB??EG8EG##########

(see usegalaxy.org to convert or use one of existing conversion scripts)
Greetings and please correct me, if i'm wrong!

[kmcarr]

I would suggest that your quality data is encoded in Illumina V1.3+, I think the fastx toolkit needs sanger encoded quality values..

Galaxy Groomer turned your sample read into:


(see usegalaxy.org to convert or use one of existing conversion scripts)
Greetings and please correct me, if i'm wrong!

Actually, the FASTX tools expect the quality values to be Illumina 1.3+ (Phred+64). If you are using Sanger encoded quality scores you would need to use the '-Q 33' option.

But I'm afraid I don't have an answer for rubi.

[rubi]

Thanks,

Actually it doesn't matter which format I'm using (Illumina or Sanger) fastx gives the same result - none of the reads are filtered out

[pbluescript]

rubi, how are you running fastx? If you are doing it on the command line, could you share your command?

[rubi]

Sure. This is the unix command I used:
fastq_quality_filter -q 20 -i <input_file> -o <output_file>

[pbluescript]

You also have to specify the percentage of bases that have a quality of q with the -p option.

[kmcarr]

You also have to specify the percentage of bases that have a quality of q with the -p option.

If you don't specify it fastq_quality trimmer uses a default of 10%.

[Apexy]

Hi,
I gave same command which was ok on one my fastq files. However the same command on another fastq file gave me this error;

fastq_quality_filter: failed to open input file 'file.fq': Value too large for defined data type

A similar error is observed for any of the commands in fastx_tool
I'll appreciate any input.
Many thanks