kmers --- generate kmer frequency tables from text streams
author: Erik Garrison <erik.garrison@bc.edu>
license: MIT
usage: ./kmers [kmermin] [kmermax]
Generate kmer frequency statistics for all kmers in stdin between
kmermin and kmermax size.
overview:
We could generate statistics for all 4-mers in the string ATGCATGC using the
following invocation.
% echo ATGCATGC | tr -d '\n' | ./kmers 4 4
kmer count freq lfreq
ATGC 2 0.4 -0.39794
CATG 1 0.2 -0.69897
GCAT 1 0.2 -0.69897
TGCA 1 0.2 -0.69897
Here count is the number of occurances, freq is the frequency of the kmer out
of all possible kmers in the input of that length (e.g. here there are 5
possible kmers of length 4 in a string of 8 characters), and lfreq is
log10(freq).
This program can be used to generate kmer frequency information for standard
FASTA reference files as follows:
% cat reference.fasta \
| sed "/^>.*$/d" \ # delete sequence names
| tr -d '\n' \ # remove newlines
| ./kmers 1 11 \ # generate kmer frequency information from 1bp to 11bp
| grep -v N \ # remove kmers containing "N"s
>reference.kmers.tsv
Note that memory usage could be extremely high with longer kmers, in the order
of kmer length * input size for each kmer length between kmermin and kmermax.