generate_profiles: Generate Protein k-mer frequency profiles

In armenabnousi/naddaR: Prediction of Protein Conserved Regions Using NADDA Algorithm

Description

constructs a dataframe where each row corresponds to one index of one protein sequence from the input dataset. It can be used to generate training and test sets to train a NADDA classification model or to predict the conserved indices of input sequences based on a trained model.

Usage

generate_profiles(obj, klen = 6, parallel = TRUE, nproc = ifelse(parallel,
  pbdMPI::comm.size(), 1), normalize = TRUE, impute = TRUE, winlen = 20,
  imputing_length = winlen%/%2, distributed = FALSE)

Arguments

obj	A filepath to a fasta file containing protein sequences or an AAStringSet object containing the sequences
klen	length of the k-mers to be used
parallel	Indicating whether the operation should be performed in parallel
nproc	Currently not supported. Will use all processors available to the job on cluster
normalize	A boolean value, indicating whether the k-mer frequencies should be normalized
impute	A boolean value, indicating whether imputed values should be inserted at the beginning and the end of the profiles
winlen	An integer, size the window used for generation of each instance
imputing_length	An integer, number of frequencies from the beginning and end of a sequence profile that should be used to impute the new values
distributed	A boolean, indicating whether the data is spread among multiple processors.

Details

If parallel is set to TRUE and distributed is set to FALSE, the method distributes the data between different processors and sets distributed to TRUE. Otherwise, if the parallel is set to FALSE and distributed is set to TRUE, the kmer frequencies are computed on each processor separately but then communicated between each other, and therefore at the end all processors have the same set of frequencies for kmers stored, using which they will generate frequency profiles for their chunk of sequences. If you prefer to run the operation in serial, set both parallel and distributed to FALSE.

Value

Returns a list with one vector for each protein sequence in the dataset. A vector for sequence s contains |s| - klen + 1 indices if impute is set to FALSE (where |s| is the length of the sequence). Otherwise it will include one index for each position in the sequence but also winlen %\% 2 indices at the beginning and end of each sequence.

Author(s)

Armen Abnousi

Examples

library(Biostrings)
library(data.table)
## Generate a set of three example protein sequences
seqs <- AAStringSet(c("seq1"="MLVVD",
                      "seq2"="PVVRA",
                      "seq3"="LVVR"))
## Count the kmers and generate a dataframe of the frequencies
profs <- generate_profiles(seqs, klen = 3, parallel = FALSE, winlen = 5, normalize = FALSE)
head(profs)
profs
##[[1]]
##[[1]]$freqs
##[1] 1.5 1.5 1.0 2.0 1.0 1.5 1.5 1.5 1.5
##[[1]]$seq
##[1] "seq1"
##
##[[2]]
##[[2]]$freqs
##[1] 1.5 1.5 1.0 2.0 1.0 1.5 1.5 1.5 1.5
##[[2]]$seq
##[[1]] "seq2"
##
##[[3]]
##[[3]]$freqs
##[1] 2 2 2 2 2 2 2 2 
##[[3]]$seq
##[1] "seq3"

armenabnousi/naddaR documentation built on May 24, 2019, 8:47 p.m.