GoMrLBP

README

Multi-resolution local binary patterns in Go

This package allows you to compute local binary pattern histograms from sequence data. Inspired by Kouchaki et al. (2019).

GoMrLBP carries out several functions to transform assembled contigs into local binary pattern histograms:

• Integer representation of sequences.
• Calculation of LBP codes.
• Construction of LBP histogram.
• Singular value decomposition of LBP histograms.

Usage example:

gomrlbp -n NUM_PROCESSORS -file CONTIG_FILE -o OUTPUT_FILE -max-win MAX WINDOW LBP -single USE SINGLE WINDOW

This implementation is an improvement over the C++ implementation in that all operations (except SVD) are carried out in parallel, allowing genomes/transcriptomes to be processed rapidly.

---

INPUTS: Contig fasta file (Can be gzipped).

OUTPUTS: Parquet file with:

• Sequence header (read ID)
• LBP histogram
• SVD results.

This package makes extensive use of gonum (https://github.com/gonum), gorgonia (https://github.com/gorgonia/gorgonia), and the truncated SVD implementation from James Bowman's nlp package (https://github.com/james-bowman/nlp).