fastx: fastq_quality_filter

[daiello]

Hello,
I used fastq_quality_filter to remove reads that had quality scores below 30 from a fastq file. The encoding of the quality scores, according to fastqc, is "Sanger / Illumina 1.9". The command I used was: fastq_quality_filter -Q 33 -q 30 -p 100 -i in.fastq -o out.fastq. I ran fastqc before and after the filtering. After filtering with fastq_quality_filter, fastqc reports that the quality encoding changed to Illumina < 1.3. Furthermore, the range of quality scores changes from 28-40 to ~4-16. According to fastqc, fastq_quality_filter is converting the original Sanger/lumina 1.9 encoding to illumina <1.3 encoding. How can I be sure fastq_quality_filter is removing the correct reads? I've attached png's of fastqc output before and after fastq_quality_filtering. Thanks a bunch.





example read from fastq file:
@HWI-ST570:48:C0NUKACXX:3:1308:18173:180116 1:N:0:
CGAGCTGGTGCGAACCAAGACCTTGGTGAAGAACAGCATCGTGGTCATCG
+
CCCFFFFFHHHHHIJJJJJJJJJJJJHIJJJJJGIIIJIJJIJJHIJJJJ

[lovenlong]

I met the same problem!

Hello,
I used fastq_quality_filter to remove reads that had quality scores below 30 from a fastq file. The encoding of the quality scores, according to fastqc, is "Sanger / Illumina 1.9". The command I used was: fastq_quality_filter -Q 33 -q 30 -p 100 -i in.fastq -o out.fastq. I ran fastqc before and after the filtering. After filtering with fastq_quality_filter, fastqc reports that the quality encoding changed to Illumina < 1.3. Furthermore, the range of quality scores changes from 28-40 to ~4-16. According to fastqc, fastq_quality_filter is converting the original Sanger/lumina 1.9 encoding to illumina <1.3 encoding. How can I be sure fastq_quality_filter is removing the correct reads? I've attached png's of fastqc output before and after fastq_quality_filtering. Thanks a bunch.
[ATTACH]1517[/ATTACH]

[ATTACH]1518[/ATTACH]


example read from fastq file:
@HWI-ST570:48:C0NUKACXX:3:1308:18173:180116 1:N:0:
CGAGCTGGTGCGAACCAAGACCTTGGTGAAGAACAGCATCGTGGTCATCG
+
CCCFFFFFHHHHHIJJJJJJJJJJJJHIJJJJJGIIIJIJJIJJHIJJJJ

Hi daiello,
Have you find the possible reason and solution for this problem ?
Whether this affects the outcome of next steps such as mapping or de novo assembly?
I met the same problem too, and I'm wondering if it is an error of fastqc but not association with the Fastq_quality_filter ?

[JackieBadger]

Without directly answering your question, I have found FASTQ to be finicky and glitchy...
Use PRINSEQ either online or from the terminal

[sisch]

This thread was almost a year old. You could have started a new one.

Still, for the sake of some answer. From the information provided, I would say it is just the FastQC software misinterpreting the quality scores. In my experience fastq_quality_filter does not convert quality scores into a different encoding. But what could happen if you discard all qualities below thirty (by chance even all below 33), is that FastQC won't find out of range quality scores and therefore misinterprets them as Illumina 1.3 instead of 1.9.

Therefore don't bother in your downstream application and have faith that the quality offset stays the same

If you need the graphs of FastQC there is most probably a setting to enforce -Q33.

EDIT: misread the date, was not "over a year old", but "almost a year old"

[lovenlong]

This thread was almost a year old. You could have started a new one.

Still, for the sake of some answer. From the information provided, I would say it is just the FastQC software misinterpreting the quality scores. In my experience fastq_quality_filter does not convert quality scores into a different encoding. But what could happen if you discard all qualities below thirty (by chance even all below 33), is that FastQC won't find out of range quality scores and therefore misinterprets them as Illumina 1.3 instead of 1.9.

Therefore don't bother in your downstream application and have faith that the quality offset stays the same

If you need the graphs of FastQC there is most probably a setting to enforce -Q33.

EDIT: misread the date, was not "over a year old", but "almost a year old"

Thanks for your suggestion.