Question regarding polishing performance in telomeric repeat regions using NextPolish2

[majssssa]

I hope this message finds you well.

I am currently using NextPolish2 to polish a high-quality genome assembly of a species in which the telomeric repeat unit is AGGGTT. During manual inspection using IGV, I observed that in several telomeric regions (highlighted in red), there appear to be two single-base errors within the repeat array. Specifically, these substitutions convert A → T and T → A, resulting in disrupted telomeric repeat units.

However, after applying NextPolish2, these apparent errors were not corrected. In addition, I also observe occasional insertion and deletion errors within the same telomeric regions.

I would like to ask for your advice on this issue. In particular:

Could this lack of correction be caused by insufficient read depth in telomeric regions or ambiguity in read mapping within highly repetitive sequences?
Are telomeric repeats generally expected to be poorly resolved by polishing tools such as NextPolish2 due to their repetitive nature and low mappability?
In such cases, do you recommend any specific parameter adjustments or alternative strategies to improve polishing accuracy in telomeric regions?

I would greatly appreciate any suggestions you may have regarding handling single-base substitutions as well as small indels in telomeric repeat regions.

Thank you very much for your time and for developing this useful tool.

Best regards,
Junpeng Ma

[moold]

As you can see in IGV, these HiFi reads differ significantly from the reference genome. However, HiFi data and assembled genomes are generally very accurate, so these reads are most likely alignment errors. NextPolish2's principle is to only correct obviously erroneous positions to prevent overcorrection.