UD Index's and amplicon length

[noah472]

Hi,

I recently received data back from our first run switching from the Nextera i5 and i7 indexes to the IDT UD indexes from Illumina and have a problem when i run make.contigs.

We have been using mothur for a long time with the old indexes and get a consistent amplicon length of ~260 bp using V4 primers.

Our new run (2 runs on separate flow cells on different days of the week sequenced using XLEAP for the first time ~Q38 quality) ends up with contigs that are 360bp.

I wrote a script that matches the 260bp from an old data set to the same sequences in the new data and found the position of the 260bp to start after 50 bases and end 50 bases before the end of the sequence.

Looking at what these 50 bases are I can see the adapter sequences and the index sequences and few extra bases at the start and end.

I tried using trimoverlap=T but this only takes the extra bases off not the adapters and the contigs end up still at 360bp. I have tried to get cutadapt to trim before processing the files but cant seem to get that to work correctly.

Does anyone have any ideas why mothur used to process these quite happily and now I have these extra sequences in, is it something to do with the UD indexes? There are a lot of polymer sequences (as you would expect) when these arent removed.

Thanks!

[pschloss]

Is it possible you’re using PE 300 reads rather than 250 PE reads?
Pat

[noah472]

Hi Pat, I did wonder that but both runs were done with PE300, the only difference on the run was XLEAP which increased the quality.

Thanks

[pschloss]

So the runs were done with PE300? If so, that’s the problem.

[noah472]

Ah, whats strange is that the first runs with the i7/i5 indexes run absolutely fine through make.contigs, its when using the UD indexes that it stops working. Is that due to the increased diversity of these sequences/indexes?

[pschloss]

Not sure what’s going on. Are you sure you were using PE300 previously? We (and Illumina I thought) strongly discourage using reads that are longer than what you are trying to sequence since it will increase the error rates.

Pat