a question about FastQC's quality scores

[kshen]

I have short reads from Illumia hiseq2000 (50bp, paired-end, DNAseq). I attached the figure of the distribution of quality scores from FastQC. It looks that the first 13 bases' quality scores are the same. So is this normal? I appreciate your help.

[danwiththeplan]

Every FastQC run I have done showed a distribution of quality scores at all points, even if it's a small distribution. So it looks weird to me.
You should probably check the scores. Make sure that they are not from an old pipeline (e.g. Illumina 1.3) since the encoding of quality scores was different. I'm guessing they might be, since it must be quite old data if it's 50bp reads. Otherwise I'm not sure what's happening.

[kcchan]

There's an orange flag at the sequence length distribution, which means that not all of your reads are 50bp. Have these fastq files already been cleaned?

[simonandrews]

You could get this effect if a large proportion of your reads had the same quality score. With some of the newer HiC datasets we've occasionally seen bases where all of the percentiles fastqc plots are the same so you just get a line rather than seeing a distribution - this plot looks a little more extensive though. The more obvious reason (as kcchan suggested) is that you're looking at a filtered data file which would artificially create this situation.