Continuous MiSeq Run Failures: ITS2

[cmc1995]

Hi all,
In 2022, I sent a pilot pool of eight of my ITS2 samples to a friend, who multiplexed them with his own and another person's samples on a 2 x 250 Miseq Nano run (My samples comprised 30% of the pool). These came back with sequences, giving us the green light to move forward.

Since that small pilot run, we pooled 4 plates of samples and have continuously run into major issues. When running our samples by themselves, the Q30 drops precipitously low around cycle 100 or so.




First, we thought that there might be adapter dimer in our pool (we saw adapter readthrough in the undetermined sequences, and there was seemingly an A overcall), despite a good tape station output, so we incorporated 3 fairly stringent Ampure bead cleans through the process (PCR I, indexing PCR, pooling). Then, when the same thing happened again, we thought maybe there was an issue with base diversity, thinking back to "what worked" for the pilot. This past week, we ran the samples on a 250x2 Miseq Nano run, multiplexed with a WGS library (50/50 split). The Q30 did not drop this time (stayed at 90%!), however, receiving the data, nearly all of the sequences are in the “undetermined” folder. Frustratingly, these sequences do match plants we were expecting for the samples, but clearly they are not assigning. Our core allowed for one base difference in the barcodes. There are far, far more undetermined barcodes (thousands) than what was in our samples, likely indicating incorrect base assignment. However, the average Q30 for these barcodes were well above 30.

Some other major differences in our pilot vs the current run:

1. We are using custom IDT indexing primers (they have been used with success before, including the other samples in the pilot run) while for our samples, we used Nextera XT primers.
2. We are using a different core.
3. The run had a lower than expected cluster density, but since this is the third time this has happened, I don’t know that it is the issue.

All in all, I am wondering if you all have any thoughts on what is going wrong here. My immediate thought is to go back to the Nextera XT, but who knows. I just don't know a way forward and my PhD somewhat hinges on this.

So, so sorry for this long post and if anything I said here sounds off (I am not a DNA master/sequencing person by a long stretch). Please let me know if there's anything else I should include (images, outputs) and I will ask the core for those. Thank you for your time and attention.

[Tom Barker]

Hi cmc1995, Illumina does recommend increasing the base diversity of amplicons by spiking in their phiX control or another diverse library. If your earlier runs with poor quality were 100% amplicon then I would expect the quality of the run to be low. We would spike in 20% phiX into our amplicon runs and it worked quite well for us so you might be able to reduce the % WGS library in your pool in future and still get good quality data. Regarding the high number of undetermined barcodes has the core facility provided you with a list of the barcodes so you can compare to the list you were expecting to see? When we have an issue like this it's usually that the i5 index sequence has been entered in the wrong orientation and it needs to be reverse complemented. if this is the case you can check some of the undetermined barcodes with the highest number of reads and see if it matches the reverse complement of what you were expecting. If it does then you can ask for it be demultiplexed again with the reverse complementary sequence and it will hopefully solve your problem.