Thread by @rrwick

I'm releasing a new tool today: Trycycler!
github.com/rrwick/Trycycl…

It is for generating a consensus long-read assembly of a bacterial genome.

(1/9)

I.e. you give Trycycler multiple different long-read assemblies of the same genome, and it produces a single consensus assembly that is better than any of the inputs.

(2/9)

In doing so, Trycycler can repair most of the problems that hide in long-read assemblies. These include:
1) missing/spurious contigs
2) bad circularisation
3) glitchy sequence regions

(3/9)

After running Trycycler, the only errors you should be left with are small-scale, e.g. homopolymer-length errors. These are from systematic basecalling errors and are to some degree unavoidable in long-read-only assemblies.

(4/9)

Polishing tools (e.g. Medaka and Pilon) can then clean up these residual small-scale errors. Therefore, given a nice hybrid (Nanopore+Illumina) read set, a Trycycler+Medaka+Pilon approach can yield an extremely high-quality genome assembly!

(5/9)

Trycycler requires some human interaction and judgement calls, which is both a good and a bad thing. It's good because it lets you clearly see when things aren't going well, e.g. if your long-read set is insufficient.

(6/9)

It's bad because it makes Trycycler not great for high-throughput assembly, i.e. it's not a good tool for assembling tons and tons of bacterial genomes.

(7/9)

Trycycler is instead a tool for taking assemblies and getting them as good as possible. It's ideal for making nice reference genomes!

(8/9)

Check out the Trycycler docs for loads more information:
github.com/rrwick/Trycycl…

(9/9)