Dear All,
Would you suggest to maintain or remove duplicate reads when studying viral subpopulation (e.g. SNP) by NGS?
I think that both approaches can bias the results...but I can not figure out which would be to most robust strategy.
If any reference is available would be perfect!
Thanks for any adivce
Giovanni
aos always depends. Do you treat your data more like genomic or transcriptomic set. Treating it like a genomic set would mean, that there the representation of a biological subtype has more or less the same chance of being sequenced, hence, removing the duplicates would actually lower the bias caused by the duplicates. If you choose to treat it more like a transcriptomic set, then you note the fact, that some parts of the data set are much less likely to be represented, thus removing the duplicates increases the bias caused by the duplicates. Your data in my understanding is actually the second type. However, I do not know the delicate intricacies of working with viral quasispecies
Since viral genomes are small chance is high that two reads are likely of different biological templates rather than PCR duplicates. So I see two options;
1) Use a PCR-free library method, if you've got enough DNA (I absolutely have no idea)
2) Use a library prep kit that uses Unique molecular identifiers (https://www.illumina.com/science/sequencing-method-explorer/kits-and-arrays/umi.html)
- Badges
- Science method