Recently we sent a batch of 24 samples of total RNA to a company for sequencing on a HiSeq 3000 platform. Both our own and the company's quality checks on the samples showed good integrity of RNA and very good purity, so they went ahead with library preparation and sequencing. After the wait we got a hard drive with the data. However, after running FastQC some of the plots show problematic issues which I'm hesitant to attribute to the samples, since they seem more of an issue with the flowcell and/or library preparation. The samples were sequenced for a downstream de-novo transcriptome assembly with Trinity, but I'm not sure if the sequences as they are now will be good enough for that, even after trimming off the adapter. I've attached the FastQC of two representative samples (for both forward and reverse reads, simplified file names). They were sequenced in different flow cells since all the samples were not sequenced together to have enough reads per sample when multiplexing. The most alarming things:
- Poor per-tile quality. Some regions of the flow cells seem like they failed in the later cycles, and they are localized, as if something had failed in one particular spot and not a generalized problem. You can also see this impacting the quality per sequence plot, where there's a hump in the sample with the worse per-tile quality.
-Adapter content. In some samples adapter starts showing up at around cycle 100, which to me suggests the fragmentation was a bit too aggressive and small fragments were used in the library preparation.
We paid quite a bit of money to have this sequenced, and it doesn't feel like the run was up to standard. Should we go back and ask the company to re-do this?
- Poor per-tile quality. Some regions of the flow cells seem like they failed in the later cycles, and they are localized, as if something had failed in one particular spot and not a generalized problem. You can also see this impacting the quality per sequence plot, where there's a hump in the sample with the worse per-tile quality.
-Adapter content. In some samples adapter starts showing up at around cycle 100, which to me suggests the fragmentation was a bit too aggressive and small fragments were used in the library preparation.
We paid quite a bit of money to have this sequenced, and it doesn't feel like the run was up to standard. Should we go back and ask the company to re-do this?
Comment