Fastq files with very high per base sequencing quality score

Question

[Ivan S]

Hello,

I am currently working with Fastq files of exome sequencing with a coverage of 150x. After running FastQC tool on these files I observe quite high Quality Score values (~35 on average) with very narrow distribution across all positions. This seems a little suspicious to me. Since I have very little experience on this type of data I'd like to ask, Is it normal to observe such high Quality Score results??

Thank you for your help

[igor]

That is normal. You can even see that in the example FastQC report: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/good_sequence_short_fastqc.html

[Ivan S]

Thanks a lot, I hadn't noticed this same tendency in the example report

[Brian Bushnell]

You can analyze the quality scores empirically if you want, via mapping; BBMap has several options for that:

bbmap.sh ref=hg19.fa in=reads.fq.gz mhist=mhist.txt qahist=qahist.txt qhist=qhist.txt

mhist generates a histogram of matches and mismatches by base position; qhist gives claimed and measured quality per position; and qahist gives the quality-score accuracy (claimed versus observed). Sometimes the quality scores are quite accurate, sometimes not; it depends on a lot of factors including luck. But if you suspect they are wrong, it's nice to validate that.

Note that humans, being diploid with a roughly 1/1000 SNP rate, have a noise floor of around 30dB for these testing methods - they work better on haploids. But they will still be fairly accurate up to Q30.

[GenoMax]

That is not surprising, if the libraries are of good quality (and the read length is not > 150).

[John]

Suspicious data you say? ಠ_ರೃ

Can we see the clues too?

[piet]

It depends on which sequencing technology you have used. If your data is from Illumina HighSeq, I would say the quality is as expected. But if your data is from Nanopore, I would also think it is suspicious.

Answer 1

[Brice Sarver]

Data I've analyzed from current sequencing platforms usually have excellent per-base quality scores. Though not always the case, I see larger 'dips' in quality scores at the beginning and end positions much less frequently than back in the earlier Illumina/454 days. You probably just have good data!