Hi all,
I sequenced a bacterial isolate using ONT with the aim of generating a final genome assembly suitable for GenBank submission. The workflow included read quality control and filtering of ONT reads, followed by assembly using Autocycler. The resulting consensus assembly was polished with Medaka, and the assembly graph was inspected in Bandage.
The Bandage visualization shows one large circular contig of approximately 3.2 Mb, along with a number of additional contigs. PlasmidFinder identifies several of the smaller contigs as plasmids, typically in the range of tens of kilobases. BUSCO analysis using the lactobacillales_odb10 / lactobacillus_odb12 datasets reports ~100% completeness for the full assembly, but with a substantial number of duplicated BUSCOs. When BUSCO is run on the chromosome-sized contig alone, completeness remains ~100%, and the all of BUSCOs are single-copy. The duplicated BUSCOs appear to be mainly contributed by one non-circular, contig.
My question concerns how to handle the remaining contigs that are not identified as plasmids by PlasmidFinder. Is it acceptable to discard such contigs if they appear to be redundant fragments (for example, prophage-related sequences, repeats, or unplaced contigs), especially when the chromosome and plasmids appear complete and circular? Or would it be better practice to analyse these contigs further before removal?
If further analysis is recommended, what would you consider a minimal and reasonable set of checks before deciding whether to discard these contigs?
Any guidance on best practices for deciding which contigs to keep versus discard in a “finished” bacterial genome submission would be greatly appreciated.
Thanks in advance!!
•
written
by
Paulina •
1 answer
One thing you can try is to run funannotate clean.
Description from Funannotate says: The script sorts contigs by size, starting with shortest contigs it uses minimap2 to find contigs duplicated elsewhere, and then removes duplicated contigs.
or you can simple self align the genome with itself using minimap2 -x asm5 and remove the small contigs which show identical alignment in one big circular contig.
Log in to answer this question.
-
hifiasm de novo assembly produces short contigs that translate to chromosomes longer than reference
written by ferran.llobet •
-
Error Scaffolding with RagTag
written by Bipin •
-
Genome Assembly Verification, Chimeric contigs
written by SomeOne •
-
How do we know that our contig sequence is really plasmid?
written by rickyalfaray •
-
Flye assembly, increased coverage results in decreased N50
written by roger.huerlimann •
-
assembly without trimming vs with trimming
written by tiagobellintani •
-
Comparing Assemblies and Breaking Inconsistent Contigs
written by EarlyEvol •
If they are real plasmids, I think it most people would agree that it would be ideal if they were kept in the finalised assembly; regardless of whether they are identified by PlasmidFinder. The non-circularised contigs (both the large one, with duplicated BUSCOs, and the small ones all with low relative coverage from Autocycler) are most likely artifacts (or at least look artifactual enough) that they could be removed.
Checking for redundancy, by aligning the all other contigs to the large complete contig, can help you with this. Although the cut-offs can then become a little arbitrary (what percentage aligning, what identity etc), in your case it might be simpler as the plasmids all have good coverage. As below
funannotate cleanis a simple way to do it. You could manually and quickly check with dgenies by aligning the assembly to itself or as above, the large complete contig vs everything else.Also, if possible I would also use a more specific/larger BUSCO dataset (I think some lactobacillales genera have their own; sometimes even a close e.g. sister genus could increase the number of BUSCOs and help).