Clonal LOHHLA results

Setup

library(ithi.utils)
load_base_libs()
library(ithi.meta)
library(ithi.clones)
library(ithi.lohhla)

library(ape)
library(phylobase)
library(phytools)
library(igraph)

clone_tree_file <- snakemake@input$clone_tree_file
clone_branch_length_file <- snakemake@input$clone_branch_length_file
clone_prevalence_file <- snakemake@input$clone_prevalence_file
clola_result_dir <- snakemake@input$clola_result_dir

db_path <- snakemake@params$db

clola_result_filenames <- list.files(file.path(clola_result_dir, "clola_condensed_results"), 
    recursive = TRUE, full.names = TRUE, pattern = "clola_results.tsv")

tree_branch_data <- read_clone_tree_data(clone_tree_file, clone_branch_length_file, 
    clone_prevalence_file, db_path)
node_mapping <- tree_node_mapping(clone_tree_file)

trees <- lapply(tree_branch_data, function(x) x$tree)

Analysis

Our goal here is to reconstruct the clonal origin of LOHHLA.

We’ve run the clonal LOHHLA (CLOLA) model in another pipeline – we’ll just visualize the results here.

This is a little preliminary still – so the LOHHLA, as called in the following plots, is pretty rudimentary. We’re just using the genotype mode for each clone – i.e. the most common genotype each clone is assigned to. In actuality, we should be doing something like checking the 95% confidence interval of allele 1/allele 2 copy number to confidently call LOHHLA.

I can quickly implement that when I get a bit of time.

## Map for CN anchoring part of model
model_map <- list(nb = "Negative binomial", mult_factor = "Multiplicative factor (lognormal)", 
    nbbin = "Negative binomial (binmedian)", student = "LogR Student t")

## Map for priors on stayrate
prior_map <- list(truncnormal = "Truncated normal", uniform = "Uniform")

for (clola_result_filename in clola_result_filenames) {
    model_type <- basename(dirname(dirname(clola_result_filename)))
    prior_type <- basename(dirname(clola_result_filename))
    
    model_type <- model_map[[model_type]]
    prior_type <- prior_map[[prior_type]]
    
    header_str <- paste("Model type: ", model_type, "; ", "Stayrate prior type: ", 
        prior_type, sep = "")
    cat("### ", header_str, "\n\n")
    
    clola_results <- fread(clola_result_filename)
    clola_results_split <- split(clola_results, f = list(clola_results$patient_id))
    clola_results_remapped <- lapply(clola_results_split, function(df) {
        patient_id <- as.character(df$patient_id[1])
        df$clone_id <- ithi.meta::df_as_map(df = node_mapping[[as.character(patient_id)]], 
            df$clone_id, from = "from", to = "to")
        labels <- trees[[patient_id]]@label
        df$clone_name <- labels[df$clone_id]
        return(df)
    }) %>% rbind.fill
    
    patients <- unique(clola_results_remapped$patient_id)
    
    for (pat in patients) {
        cat("#### Patient ", pat, "\n\n")
        
        info <- subset(clola_results_remapped, patient_id == pat)
        hla_types <- sort(unique(info$hla_locus))
        
        for (locus in hla_types) {
            cat("##### Locus ", locus, "\n\n")
            plot_tree(clola_results_remapped, trees, as.character(pat), locus, 
                colour_mode = "loh90")
        }
        cat("\n")
    }
    
}

Model type: Multiplicative factor (lognormal); Stayrate prior type: Truncated normal

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: Multiplicative factor (lognormal); Stayrate prior type: Uniform

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: Negative binomial; Stayrate prior type: Truncated normal

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: Negative binomial; Stayrate prior type: Uniform

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: Negative binomial (binmedian); Stayrate prior type: Truncated normal

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: Negative binomial (binmedian); Stayrate prior type: Uniform

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: LogR Student t; Stayrate prior type: Truncated normal

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c

Model type: LogR Student t; Stayrate prior type: Uniform

Patient 1

Locus hla_a

Locus hla_b

Locus hla_c

Patient 2

Locus hla_a

Locus hla_b

Locus hla_c

Patient 3

Locus hla_a

Locus hla_b

Locus hla_c

Patient 4

Locus hla_a

Locus hla_c

Patient 7

Locus hla_a

Locus hla_b

Locus hla_c

Patient 9

Locus hla_b

Locus hla_c

Patient 10

Locus hla_a

Locus hla_c

Patient 11

Locus hla_b

Locus hla_c

Patient 12

Locus hla_a

Locus hla_b

Locus hla_c

Patient 13

Locus hla_a

Locus hla_b

Locus hla_c

Patient 14

Locus hla_b

Locus hla_c

Patient 15

Locus hla_a

Locus hla_b

Locus hla_c

Patient 16

Locus hla_a

Locus hla_b

Locus hla_c

Patient 17

Locus hla_a

Locus hla_b

Locus hla_c