BCR phylogenetics

library(plyr)
library(dplyr)
library(data.table)
library(DT)

example_annotation_file <- snakemake@input$example_annotations
example_msa_file <- snakemake@input$example_msa

Introduction

Definitions

• Locus: sequence position (e.g. nucleotide 129)
• V(D)J recombination:
• Somatic hypermutation (SHM): Mutational process occurring within the germinal centers of secondary lymphoid organs acting primarily on BCR sequences
• Clonal family: A group/cluster of B-cells/BCR sequences related by somatic hypermutation

Parent project: ITH-Ovary

Somatic hypermutation generates sequence diversity in B-cell families

Problem

Description

Examples

Data

Example input data for a clonal family:

example_annotations <- fread(example_annotation_file)
example_annotations_info <- subset(example_annotations, select = c(id, readcount, 
    site, mut_freqs, indel_reversed_seqs))

datatable(example_annotations_info, extensions = "Buttons", options = list(pageLength = 5, 
    scrollX = TRUE, dom = "Bfrtip", buttons = c("copy", "csv", "excel", "pdf", 
        "print")))

Note that all sequences are the same length (indels have been removed).

Sequence alignment

There is considerable diversification occurring within clonal families. Evidence of this in the MSA linked includes:

• Lack of clear haplotypes (sequence positions that appear to ‘evolve together’)
• Phylogenetic patterns difficult to discern from a cursory glance at the MSA
• For each locus, mutated variant(s) are present at low frequency

Example MSA

Reading resources

Continuous time Markov chains in phylogenetics (high-ish level, sequence evolution models):

• CTMC tutorial

Basic (uniform) birth-death process:

• BDP tutorial

Branch-specific birth-death process:

• BSBDP tutorial

Literature

The concept behind this model is similar to Immunitree – see the Supplementals of the linked paper. However, the Immunitree model makes several unrealistic assumptions:

• Uniform birth-death process
• Sequence evolution occurs as a discrete process, overlaid onto the continuous-time Markov chain of the BDP
• Sequence evolution matrix assumes that the transition rate from any nucleotide to another (different) nucleotide is equal (known: transversion/transition rates are different in general, in BCRs specific codons/motifs are subject to much higher evolutionary rates)
• ‘Observed ancestor bias’: clearer to explain this in person
• Sequence error model does not incorporate read quality information

Model description

Generative process:

• “Cells” are created by a branch-specific birth-death process, rooted at a known ancestral sequence
• Sequences evolve over the branches of the BDP according to a CTMC, governed by a general time reversible (GTR) transition matrix
• Reads are produced by sequences at the tips (this could be modified to allow internal nodes to produce sequences, requires additional modeling), according to a sequence error model that incorporates read qualities