In sherrillmix/levenR: Simple Levenshtein Alignment and Distance Calculation

LevenR: Calculate Levenshtein distance between strings

Introduction

levenR provides a few functions for simple Levenshtein alignment and distance calculation with multiple threads, ends-free and reduced homopolymer gap costs.

Installation

To install directly from github, use the devtools library and run:

devtools::install_github("sherrillmix/levenR")

Examples

Generate Levenshtein distance matrix

An example of calculating the Levenshtein distance between several strings to make a distance matrix:

library(levenR)
seqs<-c('AAATA','AATA','AAAT','ACCTA')
leven(seqs)

Compare many to one

An example of calculating the Levenshtein distance between several strings against a longer reference sequence:

library(levenR)
seqs<-c('AAATA','AATA','AAAT','ACCTA')
ref<-'CCAAATACCGACC'
leven(seqs,ref,substring2=TRUE)

Find the best reference

An example of calculating the Levenshtein distance between several strings against two longer reference sequences and determining the best match for each read:

library(levenR)
seqs<-c('AAATA','AATA','AAAT','ACCTA')
refs<-c('CCATAATACCGACC','GGAAATACCTA')
dist<-leven(seqs,refs,substring2=TRUE)
apply(dist,1,which.min)

Ignoring indels in homopolymers

An example of calculating the Levenshtein distance between several strings to make a distance matrix while ignoring indels in long homopolymers (an error type common in 454 and IonTorrent sequencing):

library(levenR)
seqs<-c('AAAAATA','AAATTTTTA','AAAAATTTA')
leven(seqs,homoLimit=3)

Using multiple threads

An example of calculating the Levenshtein distance between several strings using multiple threads:

library(levenR)
seqs<-replicate(50,paste(sample(letters,100,TRUE),collapse=''))
system.time(leven(seqs))
system.time(leven(seqs,nThreads=4))

Alignment

An example of aligning strings against a longer reference:

library(levenR)
seqs<-c('AAATA','AATA','AAAT','ACCTA')
ref<-'CCAAATACCGACC'
levenAlign(seqs,ref,substring2=TRUE)

sherrillmix/levenR documentation built on Oct. 25, 2023, 11:42 a.m.