You can read those few pages online. https://github.com/rrwick/Basecalling-comparison ... Whether you are student, or a researcher, this might be of some interest to you. Not as an answer to your question, but as a source of ideas to consider.

Good luck.

You can read those few pages online. https://github.com/rrwick/Basecalling-comparison ... Whether you are student, or a researcher, this might be of some interest to you. Not as an answer to your question, but as a source of ideas to consider.

Good luck.

rrwick's github repo from Mohammed's reply would also have been my first suggestion.

The repo clearly indicates the improvements that are possible from *assembling*, i.e., creating a consensus sequence from the raw reads.

However, a "simple" assembly will only bring you so far - for now.

This means that you can actually further improve the overall accuracy by *polishing* the assembly, rather than polishing the reads *prior* to assembly.

AFAIK, there are two main ways of polishing: Using the raw reads in combination with, e.g., nanopolish, and/or using short reads, e.g., Illumina-based, with Quiver/Arrow/etc.

For an E. coli genome, we found the nanopolish step (incl. homopolymer correction and methylation-aware option) to substantially reduce the number of SNPs and, thereby, to substantially reduce the number of missed/wrongly called genes due to frameshifts (s. https://twitter.com/davidacoil/status/968917784905199616 and https://twitter.com/claczny/status/973888479359840256).

Moreover, and probably most importantly with respect to your question, running a round of polishing was able to cancel out the majority of differences between the basecallers.

In summary, it depends on what your goal is, but if assembly + polishing is an option, the basecaller has a minor impact on the sequence identity.

Best,

Cedric

this link is useful

https://github.com/rrwick/Basecalling-comparison

regards