What short read assemblers are appropriate for a tuberculosis paired fastqs coming from SRA?

Question

[conifer_genomics_uk]

What short read assemblers are appropriate for a tuberculosis paired fastqs coming from SRA? And what other tools like gap closers and scaffolders and final annotation tools to use when de novo assembling tuberculosis genome from the paired fastqs? Does anybody have similar experience and how to improve and optimize this kind of a procedure? Tried Velvet assembler and Prokka annotator. It works somehow, but far from perfect.

[GenoMax]

There are multiple M. tuberculosis genomes available at NCBI. Have you checked to see if the one you are interested in already assembled?

SPAdes is a good assembler for microbial genomes.

Answer 1

[h.mon]

It works somehow, but far from perfect.

It is not clear how good or bad are the results you are obtaining. How are you evaluating the assemblies and annotations? I would suggest you use QUAST and / or BUSCO.

For assembling bacterial genomes, I got very good results with A5_MiSeq - even better than SPAdes, but that was for an old version, never tested against current SPAdes 3.12.0.

Assembling genomes with only paired-end short read data almost always results in incomplete genomes - only genomes very small and completely devoid of repeats can potentially generate a complete assembly. You need at least mate-pairs, or even better, long reads (PacBio / NanoPore) along with short reads to get closed genomes. Even in this case, you need to "clean" your assembly afterwards, by removing very short contigs from contaminants / unassembled reads.

If all you have are the paired-end short reads, my experience is scaffolding / gap-filling don't improve the assembly at all - but I tried these approaches just a couple of times.

In addition to Prokka, you may the the RAST server for annotation.

genomax's suggestion is probably the best: check if the genome you want has been assembled.