Increase the kmer size in IDBA_UD?

Question

[mail2steff]

Dear all

Ive pair-end reads from papaya samples with average length of 151 bp. I used IDBA_UD for assembling. I increased the read length and kmer size by changing the default value as follows

static const uint32_t kMaxShortSequence = 200;

static const uint32_t kNumUint64 = 16;

./idba_ud -r reads.fa -o output mink 20 maxk 140

I have two questions.

1. When I executed IDBA_UD, It gave the contigs for the kmer size 20-100 only. I did not get any contigs for the kmer more than 100. Where I have gone wrong?
2. My second question is, The program internally decides the best kmer and produces the contig.fa file. How do we know which kmer is used to create this contig files?

Answer 1

[Sej Modha]

1. It is possible to control the min and max kmer sizes as well as the increment of the kmer steps using following parameters:

   --mink arg (=20) minimum k value (<=504)

   --maxk arg (=100) maximum k value (<=504)

   --step arg (=20) increment of k-mer of each iteration

2. If program used multiple kmer values than each output would be saved and contigs files labelled contgs.*.fa save the corresponding contigs in the fasta format. Final contigs and scaffolds are saved in the contigs.fa and scaffolds.fa files that contain the consolidated contigs and scaffolds from each kmer assembly.

[mail2steff]

Dear Sej Modha, Thank you for the commands. I increased the kmer value as above mentioned and the set the increment of k-mer of each iteration as 20. Eventhough I got the contigs for the kmwe 20, 40, 60, 80 and 100. Not more than that. Will IDBA_UD produce contigs upto 100 or AM I missing something?

With respect to the 2nd answer, How do I know which kmer is the best?