Hi,

The sequencing itself looks fine, the amount of clusters passing filter is good, also the percentage of QC30+ reads is normal. Also it does not look overloaded - in case of overload the percentage of clusters which are passing filter drops dramaticall: 90% refer to the "normal" 200 cluster density, while at 300 (150% load) it usually drops down to 50-60%. Of course, there are exceptions (when you have highly various first-read start sequences, like in scRNA-seq, for example), but in most cases when you have 15-20 samples per flow cell it is so.

If your reference is correct (genome/transcriptome, species), my guess is that the problem is in library itself. Usually when we get such low mapping rates, it is actually so. You can blast your reads agains different species - if your sample is mouse, presence of human or bacteria will proof that the reason is contamination. Anyway, I would recommend to check your library preparation protocol.

Best,

Michail

Thank you all very much for your insights!

Yurii, how do I find out the "duplication rate"? Pardon if this is a silly question. I'm a beginner. I did randomly blast some sequences against the nr database and they didn't match anything well (e

Dear Eunice, I am assuming your libraries are multiplexed, thus in reads there could be traces of adapters and primers. If you don't trim them off before aligning can lead to huge amount of unmapped reads.

All the best.

Thanks Sourav! Trimmed. Still doesn't map well...

Are you working on host-microbe system? If yes then you can consider to align reads onto host and microbe genome.

No, sponge only. Doesn't blast well with anything else...

Have you considered denovo assembly followed by gene prediction? You can then blast the predicted genes and see if there is contamination or if there is some other better reference genome.