Sequence composition

There are many instances in analyzing sequence data where you will want to know about the composition of your sequences.

For example, for a given sequence, you may want to count how many of each possible Kmer, is present in the sequence. This would be important if - for instance - you wanted to analyze the Kmer spectra of your data. Alternatively you might have a collection of sequences, and may want to count how many of each unique sequence you have in your collection. This would be important if - for instance - your collection of sequences were from a population sample, and you wanted to compute the allele or genotype frequencies for the population.

Whatever the application, BioSequences provides a method called composition, and a parametric struct called Composition to both compute, and handle the results of such sequence composition calculations.

BioSequences.Composition — Type

Sequence composition.

This is a subtype of Associative{T,Int}, and the getindex method returns the number of occurrences of a symbol or a k-mer.

source

BioSequences.composition — Function

composition(seq | kmer_iter)

Calculate composition of biological symbols in seq or k-mers in kmer_iter.

source

composition(iter)

A generalised composition algorithm, which computes the number of unique items produced by an iterable.

Example


# Example, counting unique sequences.

julia> a = dna"AAAAAAAATTTTTT"
14nt DNA Sequence:
AAAAAAAATTTTTT

julia> b = dna"AAAAAAAATTTTTT"
14nt DNA Sequence:
AAAAAAAATTTTTT

julia> c = a[5:10]
6nt DNA Sequence:
AAAATT

julia> composition([a, b, c])
Vector{BioSequences.BioSequence{BioSequences.DNAAlphabet{4}}} Composition:
  AAAATT         => 1
  AAAAAAAATTTTTT => 2

source

For example to get the nucleotide composition of a sequence:

julia> comp = composition(dna"ACGAG");

julia> comp[DNA_A]
2

julia> comp[DNA_T]
0

Composition structs behave like an associative collection, such as a Dict. But there are a few differences:

The getindex method for Composition structs is overloaded to return a default value of 0, if a key is used that is not present in the Composition.
The merge! method for two Composition structs adds counts together, unlike the merge! method for other associative containers, which would overwrite the counts.

merge! is used to accumulate composition statistics of multiple sequences:

julia> # initiaize an empty composition counter
       comp = composition(dna"");
ERROR: LoadError: UndefVarError: @dna_str not defined
in expression starting at none:2

julia> # iterate over sequences and accumulate composition statistics into `comp`
       for seq in seqs
           merge!(comp, composition(seq))
       end
ERROR: UndefVarError: seqs not defined

julia> # or functional programming style in one line
       foldl((x, y) -> merge(x, composition(y)), composition(dna""), seqs)
ERROR: LoadError: UndefVarError: @dna_str not defined
in expression starting at none:3

composition is also applicable to a k-mer iterator:

julia> comp = composition(each(DNAMer{4}, dna"ACGT"^100));

julia> comp[DNAMer("ACGT")]
100

julia> comp[DNAMer("CGTA")]
99