assemblers choose high k-mers: what does that mean?

Question

[willnotburn]

I am co-assembling metagenomic PE reads. megahitand metaSPAdesconsistently choose the final assembly as one based on the longest k-mer (141).

What does that tell about my input data and the assembly? Any cause for joy or concern?

Answer 1

[GenoMax]

• Check this paper.
  
• Effect of k-mer size on assemblies is graphically illustrated on this page.
• KMERGENIE A tool to select k-mer size by informed analysis.

[willnotburn]

Thanks, genomax. A lot in 1st and 3rd papers is - and will likely always be - outside of my field.

But I am happy to have gleaned one thing from your second bullet point. If the best k-mer length is often 80% of read length, then the longest available k-mer (141 for megahit and 127 for metaSPAdes) would make sense for my input data, which is PE 250bp.