In the beginning of my pipeline, I just fed the paired reads (2 files) into fastp, with the default options, and assumed it would do a good job preparing the reads for the next step: alignment

But I now checked the files in FastQC and it doesn't show much difference from before the fastp processing: Some extra yellow and red signals appear. I think the end result is even worse.

fastqc doesn't analyse paired reads files together. So it might be wrong regarding the paired quality, taken as a whole.

Anything else I should use together with FastQC, to verify the output quality?

I'm aligning a bunch of poolseqs and then comparing them to find consistency in the change of snp frequencies

As requested here is one of the examples that made me ask this:

Before fastp. only one problem noted by fastqc

before fastp. only one problem noted by fastqc

After fastp. overall quality seems to have decreased

after fastp. overall quality seems to have decreased

The raw data is a pool seq from illumina.

  • 3
    $\begingroup$ Please edit your question and give more details about your data (genomic/RNA-seq, whole transcriptome/3' end) and which fastqc modules are showing problems. $\endgroup$
    – Cloudberry
    Commented Jan 2, 2023 at 19:58
  • 1
    $\begingroup$ Hi fullmooninu a screenshot would be a good start, particularly if it sheds light on the QC. Would it be sufficient ... I'm sure @Cloudberry will comment. $\endgroup$
    – M__
    Commented Jan 3, 2023 at 17:18
  • $\begingroup$ Please give more context around your problem. For example, what are you trying to do with your pipeline and fastp? $\endgroup$
    – gringer
    Commented Jan 3, 2023 at 18:43
  • $\begingroup$ updated the question with your requests. $\endgroup$
    – gl00ten
    Commented Jan 3, 2023 at 23:33
  • 1
    $\begingroup$ What makes you think that quality of reads went down after running fastp? You have two warnings after running fastp. One is GC content, and other is length distribution. Both of them can be easily explained by the removal of adapter and low quality reads. You should also read fastp manual and see if running the default is a good idea. $\endgroup$
    – Supertech
    Commented Jan 4, 2023 at 3:49

1 Answer 1


On a regular basis, using FastQC is quite enough as to assess the quality of FastQ-formatted data. It gives you plenty of details as can be seen in the screenshots you shared.

As others have commented or suggested, there is nothing wrong with having more 'yellow' warnings after processing the FASTQ files using fastp. At lest not as long as what you observe in the data looks fine and the warning has some explanation.

The sequence length distribution warning is very often present post-processing since at the begging all your reads are of the same length but this is not the case any more after you trim the adapters and the low quality bases at the end of the read for example. You have to go to the plots specific to that section and make sure all looks fine. It should be the case.

The GC sequence content warning. It depends a lot what organism you are sequencing, which regions of the genomes, etc. Importantly, what is the expected GC content of those regions. So you can compare to these observed numbers. Once again the warning can be easily triggered after the processing, since for a start Illumina sequencing has known GC bias

Finally, I think actually your reads look better after the processing with fastp. You are showing us the per base sequence quality and before, there are many reads with bases on the yellow and even the red sections of the quality scale. But this is not the case any more after the processing !

  • $\begingroup$ Thanks. I still suspect a tool like fasqc, pair ended aware, will give better results. $\endgroup$
    – gl00ten
    Commented Jan 10, 2023 at 10:37
  • $\begingroup$ You're wrong about fastqc being PE aware as well as fastp not being PE aware. Are you even reading the manuals, @fullmooninu? $\endgroup$
    – Ram RS
    Commented Jan 10, 2023 at 16:11
  • 5
    $\begingroup$ I am always amazed by users new to the field and then calling out on of the most extensively used tools in all of bioinformatics. There is nothing wrong with your data in the first place. Quality does not go down after trimming as seen in ghe per-base boxplots, the length warning is expected due after any trim and GC content is just deviation from a theoretical distribution. There is no basis to call results worse or regarding fastqc as not accurate. $\endgroup$
    – ATpoint
    Commented Jan 10, 2023 at 17:31
  • $\begingroup$ @RamRS i didn't find out that in the manual. can you point it out for me, please? $\endgroup$
    – gl00ten
    Commented Jan 16, 2023 at 15:03
  • $\begingroup$ @fullmooninu Literally in the second usage example here: github.com/OpenGene/fastp#simple-usage $\endgroup$
    – Ram RS
    Commented Jan 18, 2023 at 17:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.