The purpose of this package is to describe the BLAST alignments among a set of peptide sequences by reporting the overlaps of each peptide’s alignments to other peptides in the set. One can imagine inputting a list of peptides enriched by immunoprecipitation (e.g. by PhIP-seq) to identify corresponding epitopes.

epitopefindr takes a .fasta file listing peptide sequences of interest and calls BLASTp from within R to identify alignments among these peptides. Each peptide’s alignments to other peptides are then simplified to the minimal number of “non overlapping” intervals* of the index peptide that represent all alignments to other peptides reported by BLAST. (*By default, each interval must be at least 7 amino acids long, and two intervals are considered NOT overlapping if they share 6 or fewer amino acids). After the minimal overlaps are identified for each peptide, these overlaps are gathered into aligning groups based on the initial BLAST. For each group, a multiple sequence alignment logo (motif) is generated to represent the collective sequence. Additionally, a spreadsheet is written to list the final trimmed amino acid sequences and some metadata.

workflow

Installation:

  1. Install R (version 3.5+).
  2. Install BLAST+ (version 2.7.1). (Note: we have observed some issues with more recent versions of BLAST+ and will monitor for bugfixes.)
  3. In R console, execute:
if (!requireNamespace("devtools")) install.packages("devtools")
devtools::install_github("brandonsie/epitopefindr")
library(epitopefindr)

Optional (Suggested) Additional Setup :

(These are not essential to epitopefindr, but are used to generate alignment logo PDFs from the alignment data, which can be valuable visualizations.)

  1. Install a TeX distribution with pdflatex. (e.g. MiKTeX, Tex Live). (Optional; used to convert multiple sequence alignment TeX files to PDF.)
  2. Install pdftk (version 2.02+). (Optional; used to merge individual PDFs into a single file.) If you are unable to install pdftk, but your system has the pdfunite command line utility, you can install the R package pdfuniter, which performs a similar function. With pdfuniter, run epfind with pdftk = FALSE, pdfunite = TRUE.

Debugging

  • epitopefindr 1.1.29 (2020-05-30) updates the DESCRIPTION file to specify sources of Bioconductor and Github packages. If the above installation produces issues during certain package installations, try the following:
if (!requireNamespace("BiocManager")) install.packages("BiocManager")
BiocManager::install(c("Biostrings", "IRanges", "msa", "S4Vectors"))

# Install Github packages
if(!requireNamespace("devtools")) install.packages("devtools")
devtools::install_github("mhahsler/rBLAST") 
devtools::install_github("brandonsie/pdfuniter")   
devtools::install_github("brandonsie/epitopefindr")

Guide

  1. Prepare a list of your peptides of interest using one of the following two methods. Either of these can be fed as the first input parameter to epfind.
  2. To run a typical epitopefindr pipeline, try calling epfind:
# Basic call
epfind(<path to .fasta>, <path to output dir>)

# Without pdflatex or pdftk
epfind(<path to .fasta>, <path to output dir>, 
        pdflatex = FALSE, pdftk = FALSE)

# More stringent e-value threshold
epfind(<path to .fasta>, <path to output dir>, e.thresh = 0.0001)

You can try running epfind() with some provided example data:

my_peptides <- epitopefindr::pairwise_viral_hits
epitopefindr::epfind(data = my_peptides, output.dir = "my_epf_1/")

A brief summary of the functions called by epfind:

  • pbCycleBLAST cycles through each input peptide to find the overlap of its alignment with other peptides from the input. Nested within a call to pbCycleBLAST are calls to epitopeBLAST, indexEpitopes.
  • trimEpitopes performs a second pass through the identified sequences to tidy alignments.
  • indexGroups collects trimmed sequences into aligning groups
  • groupMSA creates a multiple sequence alignment motif logo for each group
  • outputTable creates a spreadsheet summarizing identified sequences and epitope groups