Failed index reads with homegrown adapters

[RachelD]

Hi,

I'm hoping someone can help. I am trying to sequence amplicons on the MiSeq with adapters and barcodes we have designed in the lab. The same sequences have been used successfully with other targets. My sequences are ending up in undetermined reads due to an index failure but the read 1 and 2 sequences look fine. I have been pooling one or two of my libraries with Illumina kit-generated libraries on the run and the other libraries have worked. I have also Sanger sequenced the amplicons to check that the barcodes are correct.

Any suggestions?!

Thanks,
Rachel

[Genetic Librarian]

Hi Rachel,

what ended up in the undetermined bin?
Just some random homopolymers or actual index sequences?
If you look at that (e.g. in the html report), you usually can spot if you find the reverse complement or reverse indices in your undetermined bin. That is usually the no.1 thing to look for.

[RachelD]

Hi,
The actual sequence data for the insert/amplicons went into undetermined reads. The indexes/barcodes did not sequence. I've checked all the indexes/barcodes that were read on the run and there is nothing looking like mine.

Thanks!

[GenoMax]

It is possible to get the index read as a separate fastq file with bcl2fastq v.2.xx. If you have access to the raw data folder you may want to check into getting that file. If those reads look like nothing you expect (to a large extent, there will always be permutations of bases that can't be explained easily but that fraction should be relatively small) then there must be an error in your experimental design (but you can worry about that later).

[thermophile]

In your sample sheet have you reverse complimented your i7 sequences? i5 is sequenced 5'->3', i7 is sequenced 3'->5'

[cmbetts]

If there isn't anything sensitive, you could post your adapter/primer sequences and sample sheet and have another set of eyes make sure everything is right on that end before trying to diagnose what could have happened at sample prep and sequencing

[moatman]

It is possible to get the index read as a separate fastq file with bcl2fastq v.2.xx. If you have access to the raw data folder you may want to check into getting that file. If those reads look like nothing you expect (to a large extent, there will always be permutations of bases that can't be explained easily but that fraction should be relatively small) then there must be an error in your experimental design (but you can worry about that later).

You can also tweak a file on the MiSeq to configure it to create index reads for every sample(and Undetermined) by following the guide here: https://github.com/aryeelab/umi/wiki...ut-index-reads It's basically just adding one line to a text file telling it to output index reads.