ONT multiplexing not generating equal amount of data for each barcodes

Question

[chayan.roy93]

Hi, I have multiplexed just two samples to run on promethION. I have used Qubit and Nanodrop to check the quality and quantity of the DNA, and it's good (way above the permissible range). So I pooled them at equal quantity; 400ng in 7.5 microL; as recommended. But I am actually getting more than 10 times of data for one sample. I stopped the run, rechecked my sample and started rerunning from scratch after washing the flow shell. But the same thing is happening again. I was wondering if someone has faced this before and can help on how to fix this?

Someone suggested me that in one sample the number of modified bases might be higher and thus slowing down the rate significantly for that particular sample. How to tackle that one, I have no idea. In that case the average quality will be less for the samples generating less number of reads. Can any one suggests any other checks to determine whether it is the sole effect of modified bases in my sample or there could be other things to watch out for?

Any help is much appreciated.

[andres.firrincieli]

Hi chayan,

We could give you some advice about the base-calling or demultiplexing processes, but this sounds like something that can be solved, if possible, by the ONT support team because it seems to be an issue with library preparation or the sequencing run.

Can any one suggests any other checks to determine whether it is the sole effect of modified bases in my sample or there could be other things to watch out for?

If you still have the fast5 files of each sample you could check the methylation state of your DNA by following this protocol: Quickstart - calling methylation with nanopolish

Keep in mind that the most recent versions of guppy (6.1.1) provide different configuration files (.cfg) for the detection of modified bases during the base-calling

[samuel.a.odonnell]

My first guess: Although you said you have measured your concentration; perhaps the molarity is different between samples. I.e the fragment distribution is different. I certainly have seen that many times (where a more fragmented library will make up more of the library in terms of reads); regardless of extracting samples in parallel. Perhaps you can already see a difference in read lengths between the two samples based on your initial output?

[colindaven]

Not an answer, but it's a really good idea to check library barcode distributions on a flongle or even briefly on a minion flowcell before proceeding to a big promethion sequencer. That's how most service providers I have spoken to have done it too.