Assemble and get consensus of inserted sequence

Question

[WouterDeCoster]

I'm working on some code to improve the identification of inserted sequences (structural variant) from nanopore sequencing. Extracting all reads with insertions and extracted inserted sequences is the easy part, but the next one would be to find a suitable algorithm to assemble those and get a consensus sequence. I feel like going for a genome assembler such as Canu would be overkill for a job like this.

So I have for each position about 20-40 sequences of at least 50bp (and up to say ~10000 and rarely much more), derived from nanopore sequencing so they're rather noisy on the single nucleotide level. Which algorithm/tool would be fit for this purpose?

[GenoMax]

tadpole.sh from BBMap suite is a k-mer based assembler that may fit the bill. It can extend/error correct. A guide is available here.