End of contig extended by single read past telomere sequence

[kcamnairb]

Hi, I used Flye to assemble a fungal genome using PacBio Sequel subreads into chromosome level contigs with over 100X coverage. Over half of the contigs contain the expected telomere sequence near the ends, but there are several contigs that have sequence extended beyond the telomeres by only one read. I copied two examples below. The track above the reads shows the telomere sequence match. The sequence after the telomere seems like some kind of artifact. Is there a setting to require a minimum coverage, or do you have any ideas what might be causing this?

Thanks,
Brian

[mikolmogorov]

Hi Brian,

Definitely looks like a Flye artifact. Currently, the algorithm does require support of multiple reads within the contig, but contig ends may be extended with a single longest read (which seemed to happen in this case).

I am currently working on polishing improvements which potentially should help fixing this. In the meantime, there are a few things to try as an ad hoc solution. First, you can manually remove the offending reads from the set and reassemble (you can restart from the contigger stage). Alternatively, if you extract fasta sequences from the assembly_graph.gfa, they would contain edges before they are expanded using longest reads, so they should not have this artifacts (but possibly are a bit shorter).

Hope this helps,
Mikhail