re-mapping with quiver

[joxcargator73]

Hello
I am dealing with a high repetitive bacterial genome. After running the PBcR and polished the assembly with quiver I still have around 207 errors when I compare my assembly with my reference genome (the same genome sequenced with sanger).
I learned that re-mapping with pbaliagn and quiver could help to improve. A second re-mapping improved the number of errors (164 now).
I run a third re-mapping and the error decreased to 107. I know that the re-mapping can help to fix the errors but it is possible that I am picking new errors in the new consensus. Does anybody has tried multiple mapping for assembly polishing?.
Is it possible to obtain a certain stabilization in the number of errors after multiple polishing?
Thanks

[rhall]

Multiple rounds of resequencing and quiver can improve the base accuracy, particularly in cases where repetitive elements cause miss-mapping of reads, but I would be careful.
Looking at the alignment and trying to understand the differences in consensus calls and the sanger reference will probably go a long way in helping understand why the errors don't get correctly called in the first round of quiver. Are the errors in repetitive elements, is the mapping QV low in these regions? Are any of the base calls lowercase? (quiver calls bases as lowercase if the algorithm does not converge, of there is not enough coverage to accurately call).
Note, for an average bacterial genome, ~200 errors is > QV50.
Quiver should stabilize and stop making changes to the consensus after ~2-3 rounds, if it doesn't this would indicate that the mapping of reads to the repetitive elements is not unambiguous, if you have enough coverage, limiting the reads that are used in the mapping to a higher length may help.