low quality amplicons

[Msweetlo]

Hi all,

I recently received the results back from a MiSeq PE-300 amplicon run, and found really weird results that we cannot explain.

We sequenced amplicons of two eukaryotic marker genes originating from soil samples. Both markers were separately PCR'd and subsequently pooled, while we added technical replicates of three of these samples together with a mock community made from pooled DNA (also replicated 3 times). We received a sufficient amount of reads per samples, but upon closer inspection, the quality in some samples dropped after around 100bp to <phred 5. We found this in the forward as well as the reverse reads. This quality drop seemed to be associated with the primer pair of just one of the two marker genes, which in bad samples was overrepresented (excluding the other gene). Notably, in the technical replicates and mock communities, we see that some replicates are good, while others are very bad...

In a second remark: the bad quality amplicons match nothing with BLAST searches, and seem to have no resemblance to the targeted gene. Yet the primers picking up non-target genes seems unlikely to us since both primers are present in the amplicon, and the Bioanalyzer results of the initial PCR products show fragments of the correct length...

The feed-back we received from the company is that the run itself went good, with sufficient clusterdensity...

The only thing that went wrong is that the Nextera kit prepared library went passed its expiry date in the sequencing company (they let us wait for 2 months).

Does anyone have any ideas what could have happened? Could these weird results be explained by the expiry of the prepared library? Thanks in advance!

[matth431]

What version of MCS is the MiSeq running? We had issues with recent runs until Illumina suggested upgrading to v2.5.

We do still see issues with some samples - one user experiences a massive quality drop at base 8 with subsequent recovery - on inspection, the sequence hits a run of ~6-7 G residues around that base.

[Msweetlo]

The sequencing was done by and external company, and they are not very keen to share much information.

I must remark the quality drop we observe happens when the sequence hits a poly-A residue (with no recovery), however, the bases before this quality drop also seem to be (although highly replicated) unknown DNA.

[esherman]

How long is the amplicon that had poor quality scores? My guess is that you're sequencing the entire amplicon, through the Illumina p5/p7/primer landing sequences, and into the flow cell. I think Illumina extends the covalently-bound p5/p7 sequences off the flow cell with T's (to reduce steric issues associated with cluster generation) so when they're sequenced, they become A's. After the A's, there's no more DNA and the cluster goes dark. The machine picks up residual noise from surrounding clusters or lagging strands within the cluster and thus attempts to give the cluster a base call, but the signal to noise ratio is so low that the Q-score drops significantly. To confirm that this is the case, check the 3' sequences (before the A's) to see if it matches the Illumina p5/p7 sequences (irregardless of how your amplicon primers were designed, they need the p5/p7 sequences to cluster on the flow cell - this immensely helpful post highlights those sequences explicitly). I suspect that the highly replicated unknown DNA you're seeing is in fact Illumina p5/p7 sequence.

Remember that your bioanalyzer size gives you the length of the full PCR fragment, Illumina sequences included. Each sequencing read starts ~60-70bp into the molecule, so a 150bp fragment on the bioanalyzer will start running into A's around base pair 90 or so, and the cluster will go dark (and have atrocious Q-scores) around base pair 100.