following a conversion during a conference a few weeks ago, I used this tool to identify some misassemblies in my metaMDBG assemblies, and found some examples of weird cases where the genome is covered by 0 read (minimap2 mapping). One example here:
I compare results with hifiasm and metaMDBG shows a lot more cases (using 6 samples), however, it seems that still impacts a very low % of the total assembly (<0.1%):
Top plot is the cumulative length (bp) of regions with 0 coverage for each sample.
Funny enough, while hifiasm is doing better for 5 samples, there is one that has a lot of similar regions. However, it seems that those regions are found on the border of the assembly only, while for metaMDBG it can be found in the middle of a contig (like the screenshot).
I was wonder why you think this might happen. One could assume mapping errors but this seems a bit odd.
Hi,
Thanks for your feedback. It's not mapping issue, this is definitely missassemblies.
I'm going to check if I can fix them easily, or maybe fix them as a post-processing step.
Hi,
following a conversion during a conference a few weeks ago, I used this tool to identify some misassemblies in my metaMDBG assemblies, and found some examples of weird cases where the genome is covered by 0 read (minimap2 mapping). One example here:
I compare results with hifiasm and metaMDBG shows a lot more cases (using 6 samples), however, it seems that still impacts a very low % of the total assembly (<0.1%):
Top plot is the cumulative length (bp) of regions with 0 coverage for each sample.
Funny enough, while hifiasm is doing better for 5 samples, there is one that has a lot of similar regions. However, it seems that those regions are found on the border of the assembly only, while for metaMDBG it can be found in the middle of a contig (like the screenshot).
I was wonder why you think this might happen. One could assume mapping errors but this seems a bit odd.
Thanks, JS