Short reads - dodgy consumables?

[nickloman]

After a short spell of good results, again we're getting way short reads on our 454 Titanium WGS runs. Last time this was down to dodgy consumables and Roche put a release out admitting there was a QC problem. Anyone else having problems recently?

[Zaag]

Yes, but not very consistent so this could be due to sample problems.

[genes123]

Same for us, we've had 5-6 runs of very short reads (250bp average) in the past 3 weeks after 3-4 months of 400+bp (many samples had previously been successfully sequenced so not sample related)

[nickloman]

Ah.

Perhaps it would be useful to compare lot numbers on consumables, do you have them to hand?

[genes123]

Since we started using Seq kit 93801620 (for which they announced a new formulation of their substrate solution) although correlation is not perfect.

[nickloman]

The engineer came:

It seems some of the tubes inside were dirty and wash flow may have been insufficient which means removal of residual dNTP after each cycle would be incomplete and so as the reads progress, there is increasing corruption of the read sequence due to overlapping reads after displacement of the end of the strand 'forwards' by illegitimate insertion of bases that should not be available in the next cycle (if I've understood correctly), a phenomenon called read-forward. Apparently Roche's analysis of the data suggested that this was indeed the problem.

This suggests to us we need to make sure the tubes are replaced at no more than 12 mo intervals and possibly as little as 6 mo.

(Edited this post for clarity)