K-mer counting with Jellyfish
Help 5 / 6
Count k-mers in sequencing reads

So far, we counted k-mers on entire genomes. In this section, we will count k-mers of sequencing reads and see if we can classify them to one or the other genome.

Let’s simulate some sequencing reads from Dengue using wgsim:

wgsim \
   -N 1000 \
   dengue.fa \
   r1.fastq \
   r2.fastq

Now let’s count k-mers in the sequencing reads, but only if they are also found in the Dengue genome (we’ll ignore R2 reads):

jellyfish count \
   -m 9 \
   -s 15000 \
   -C \
   -o map_to_dengue.jf \
   --if dengue.fa \
   r1.fastq

Note that we’re using -C, which means Jellyfish will only count canonical k-mers.

For example, the k-mers ACCT and AGGT are reverse complements of each other, so their counts are grouped under the so-called canonical k-mer ACCT, because it comes first alphabetically.

Note that we did not use -C earlier when counting k-mers on genomes. The difference is: when sequencing DNA, we can’t differentiate which of the 2 strands the read comes from.

Next, let’s count k-mers from Dengue sequencing reads that are also found in the Chikungunya genome:

jellyfish count \
   -m 9 \
   -s 15000 \
   -C \
   -o map_to_chikungunya.jf \
   --if chikungunya.fa \
   r1.fastq

Looking at the distribution of k-mers, there are thousands more k-mers from Chikungunya that don’t show up in the sequencing reads:

jellyfish histo map_to_dengue.jf
jellyfish histo map_to_chikungunya.jf

This is a simplistic example but illustrates how k-mers and their distribution can be a useful way to summarize sequencing reads for classification.

← Previous Next →