Simulation Benchmarking
Last updated: 2020-04-23
Checks: 6 1
Knit directory: MINTIE_paper_analysis/
This reproducible R Markdown analysis was created with workflowr (version 1.6.1). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
The R Markdown is untracked by Git. To know which version of the R Markdown file created these results, you’ll want to first commit it to the Git repo. If you’re still working on the analysis, you can ignore this warning. When you’re finished, you can run wflow_publish to commit the R Markdown file and build the HTML.
Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
The command set.seed(20200415) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.
The results in this page were generated with repository version 9f95a7c. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
Ignored files:
Ignored: .DS_Store
Ignored: .Rhistory
Ignored: .Rproj.user/
Ignored: R/
Ignored: analysis/.DS_Store
Ignored: code/.DS_Store
Ignored: data/.DS_Store
Untracked files:
Untracked: MINTIE_paper_analysis.Rproj
Untracked: analysis/Simulation_Benchmarking.Rmd
Untracked: analysis/cache/
Untracked: code/plot.R
Untracked: code/simu_helper.R
Untracked: data/simu/
Unstaged changes:
Modified: .Rprofile
Deleted: MINTIE_Rmarkdown.Rproj
Modified: README.md
Modified: analysis/_site.yml
Deleted: analysis/about.Rmd
Modified: analysis/index.Rmd
Deleted: code/README.md
Deleted: data/README.md
Staged changes:
Modified: .gitignore
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
There are no past versions. Publish this analysis with wflow_publish() to start tracking its development.
# util
library(data.table)
library(dplyr)
library(here)
# plotting
library(ggplot2)
# bioinformatics helpers
library(GenomicRanges)
library(Rsamtools)source(here("code/simu_helper.R"))
source(here("code/plot.R"))options(stringsAsFactors = FALSE)Simulation Benchmarking
Here we analyse the results on a simulated data set of 1,500 variants containing fusions, transcribed structural variants and novel splice variants, across 8 methods: MINTIE, TAP, Barnacle, SQUID, JAFFA, Arriba, StringTie and KisSplice.
The plots show the number of variants detected from each category and the number of false positives. The number of background genes detected in the false positives in displayed as printed output.
fus_truth <- read.delim(here("data/simu/truth/allvars_fusions_simulated.tsv"))
tsv_nsv_truth <- read.delim(here("data/simu/truth/allvars_tsvs_splice_simulated.tsv"))
all_gene_locs <- read.delim(gzfile(here("data/simu/all_gene_locs.tsv.gz")))
bg_gene_ref <- read.delim(here("data/simu/truth/bg_gene_ref.tsv"), header = TRUE)
# extract truth and background gene names
var_genes_truth <- unique(c(fus_truth$gene1, fus_truth$gene2, tsv_nsv_truth$gene))
var_genes_truth <- var_genes_truth[var_genes_truth != ""]
bg_genes <- bg_gene_ref$gene
# make Genomic Ranges objects from data
fus_grx <- get_granges(fus_truth)
tsv_nsv_grx <- get_granges(tsv_nsv_truth)
bgenes_grx <- get_granges(bg_gene_ref, convert_chrom = TRUE)
all_gene_grx <- get_granges(all_gene_locs, convert_chrom = TRUE, add_chr = FALSE)
# create results object
nsv_names <- c("Extended exon",
"Novel exon",
"Retained intron",
"Truncated exon",
"Unknown splice")
tsv_names <- c("Deletion",
"Insertion",
"Internal tandem duplication",
"Partial tandem duplication",
"Inversion")
fus_names <- c("Fusion (canonical)",
"Fusion (EE)",
"Fusion (NE)",
"Fusion (insertion)",
"Fusion (unpartnered)")
vartypes <- c("False positive", nsv_names, tsv_names, fus_names)
classes <- c("False positive",
"Novel splice variant",
"Transcribed structural variant",
"Fusion")
results <- data.frame(vartype = factor(vartypes, levels = vartypes),
class = factor(c(classes[1], rep(classes[2:4], each=5)),
levels = classes[4:1]),
row.names = c("FP", "EE", "NE", "RI", "NEJ", "US",
"DEL", "INS", "ITD", "PTD", "INV",
"canonical_fusion", "EE_fusion",
"NE_fusion", "INS_fusion", "unpartnered_fusion"))
cols <- c("#87649aff", "#bdd888ff", "#e7d992ff", "#636363")
names(cols) <- c("Fusion", "Transcribed structural variant", "Novel splice variant", "False positive")MINTIE
# load data
mintie_results <- read.delim(here("data/simu/results/MINTIE/allvars-case_results.tsv"))
mintie_results <- mintie_results[mintie_results$logFC > 5,]
# extract genes
mintie_vargenes <- sapply(mintie_results$overlapping_genes, function(x){strsplit(x, "\\||:")[[1]]})
mvg <- unlist(mintie_vargenes)
# count found fusions, TSVs and NSVs
found_fus <- table(fus_truth$fusion_type[fus_truth$gene1 %in% mvg | fus_truth$gene2 %in% mvg])
found_tsv <- table(tsv_nsv_truth$vartype[tsv_nsv_truth$gene %in% mvg])
# count false positives genes
fp <- sapply(mintie_vargenes, function(x){!any(x %in% var_genes_truth)})
fp_genes <- unlist(sapply(mintie_results$overlapping_genes[fp], function(x){strsplit(x, "\\||:")}))
fp_genes <- fp_genes[fp_genes != ""]
n_fp <- length(unique(fp_genes))
results <- append_results(results, "MINTIE", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(fp_genes %in% bg_genes)))[1] "Background genes in FPs: 3"plot_simu_benchmarking(results, "MINTIE")
TAP
# load result data
tap_svs <- read.delim(here("data/simu/results/TAP/sv.bedpe"), sep = "\t", skip = 2)
tap_nsv <- read.delim(here("data/simu/results/TAP/novel_splicing.bedpe"), sep = "\t", skip = 2)
# extract variant genes
tap_vargenes <- Reduce(union, list(tap_svs$gene1, tap_svs$gene2, tap_nsv$gene))
# count found fusions, TSVs and NSVs
found_fus <- table(fus_truth$fusion_type[fus_truth$gene1 %in% tap_vargenes | fus_truth$gene2 %in% tap_vargenes])
found_tsv <- table(tsv_nsv_truth$vartype[tsv_nsv_truth$gene %in% tap_vargenes])
# count false positives genes
fp_svs <- tap_svs[!tap_svs$gene1 %in% var_genes_truth & !tap_svs$gene2 %in% var_genes_truth,]
fp_nsv <- tap_svs[!tap_svs$gene %in% var_genes_truth,]
n_fp <- nrow(fp_nsv)
results <- append_results(results, "TAP", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(fp_genes %in% bg_genes)))[1] "Background genes in FPs: 3"plot_simu_benchmarking(results, "TAP")
Barnacle
# load hg19 truth locations and make Genomic Ranges objects
hg19_fus1_grx <- get_granges_from_bed(here("data/simu/truth/hg19_liftover/fus_loc1.bed"))
hg19_fus2_grx <- get_granges_from_bed(here("data/simu/truth/hg19_liftover/fus_loc2.bed"))
hg19_tsv_nsv_grx <- get_granges_from_bed(here("data/simu/truth/hg19_liftover/tsv_nsv.bed"))
hg19_bgenes_grx <- get_granges_from_bed(here("data/simu/truth/hg19_liftover/background_genes.bed"))
# load barnacle results and make Genomic Ranges objects
barnacle_results <- read.delim(here("data/simu/results/barnacle/allvars-case.barnacle.data"), header=FALSE)$V1
barnacle_results <- barnacle_results[grep("OVERLAPPING", barnacle_results)]
barnacle_grx_a <- get_barnacle_grx(barnacle_results)
barnacle_grx_b <- get_barnacle_grx(barnacle_results, side_A = FALSE)
# extract hits
hits <- get_hits(barnacle_grx_a, barnacle_grx_b,
hg19_fus1_grx, hg19_fus2_grx,
hg19_tsv_nsv_grx, hg19_bgenes_grx)
found_fus <- table(fus_truth[hits$fus_truth_hits,]$fusion_type)
found_tsv <- table(tsv_nsv_truth[hits$tsv_nsv_truth_hits,]$vartype)
n_fp <- sum(!1:length(barnacle_grx_a) %in% unique(hits$caller_hits))
results <- append_results(results, "Barnacle", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(hits$bg_hits %in% hits$caller_hits)))[1] "Background genes in FPs: 10"plot_simu_benchmarking(results, "Barnacle")
SQUID
# load data
squid_results <- read.delim(here("data/simu/results/squid/results_sv.txt"))
# make GRanges objects
chr1 <- sapply(squid_results$X..chrom1, convert_chrom, add_chr=FALSE)
loc1_grx <- GRanges(seqnames = chr1,
ranges = IRanges(start = squid_results$start1, end = squid_results$end1))
chr2 <- sapply(squid_results$chrom2, convert_chrom, add_chr=FALSE)
loc2_grx <- GRanges(seqnames = chr2,
ranges = IRanges(start = squid_results$start2, end = squid_results$end2))
# get hits
hits <- get_hits(loc1_grx, loc2_grx, fus_grx[[1]], fus_grx[[2]], tsv_nsv_grx, bgenes_grx)
found_fus <- table(fus_truth[hits$fus_truth_hits,]$fusion_type)
found_tsv <- table(tsv_nsv_truth[hits$tsv_nsv_truth_hits,]$vartype)
# get false positives
fps <- squid_results[!rownames(squid_results) %in% hits$caller_hits,]
fps <- fps[!(fps$X..chrom1 %like% "alt" | fps$chrom2 %like% "alt"),]
n_fp <- nrow(fps)
results <- append_results(results, "SQUID", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(hits$bg_hits %in% hits$caller_hits)))[1] "Background genes in FPs: 0"plot_simu_benchmarking(results, "SQUID")
JAFFA
# load data
jaffa_results <- read.delim(here("data/simu/results/JAFFA/jaffa_results.csv"), sep = ",")
# make GRanges objects
chr1 <- sapply(jaffa_results$chrom1, convert_chrom, add_chr=FALSE)
loc1_grx <- GRanges(seqnames = chr1,
ranges = IRanges(start = jaffa_results$base1, end = jaffa_results$base1))
chr2 <- sapply(jaffa_results$chrom1, convert_chrom, add_chr=FALSE)
loc2_grx <- GRanges(seqnames = chr2,
ranges = IRanges(start = jaffa_results$base2, end = jaffa_results$base2))
# get hits
hits <- get_hits(loc1_grx, loc2_grx, fus_grx[[1]], fus_grx[[2]], tsv_nsv_grx, bgenes_grx)
found_fus <- table(fus_truth[hits$fus_truth_hits,]$fusion_type)
found_tsv <- table(tsv_nsv_truth[hits$tsv_nsv_truth_hits,]$vartype)
# get false positives
fps <- jaffa_results[!rownames(jaffa_results) %in% hits$caller_hits,]
fps <- fps[!(fps$chrom1 %like% "alt" | fps$chrom2 %like% "alt"),]
fp_genes <- unlist(sapply(fps$fusion.genes, strsplit, split=":"))
n_fp <- length(unique(fp_genes))
results <- append_results(results, "JAFFA", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(hits$bg_hits %in% hits$caller_hits)))[1] "Background genes in FPs: 0"plot_simu_benchmarking(results, "JAFFA")
Arriba
# load data
arriba_results <- read.delim(here("data/simu/results/arriba/fusions.tsv"))
# extract variasnt genes
arriba_vargenes1 <- sapply(arriba_results$X.gene1, get_arriba_genes)
arriba_vargenes2 <- sapply(arriba_results$gene2, get_arriba_genes)
avg <- unlist(list(arriba_vargenes1, arriba_vargenes2))
# tally results
found_tsv <- table(tsv_nsv_truth[tsv_nsv_truth$gene %in% avg,]$vartype)
found_fus <- table(fus_truth[fus_truth$gene1 %in% avg | fus_truth$gene2 %in% avg,]$fusion_type)
# get false positives
fp1 <- sapply(arriba_vargenes1, function(x){!any(x %in% var_genes_truth)})
fp2 <- sapply(arriba_vargenes2, function(x){!any(x %in% var_genes_truth)})
fp_genes1 <- arriba_vargenes1[fp1 & fp2]
fp_genes2 <- arriba_vargenes2[fp1 & fp2]
fp_genes <- unique(union(fp_genes1, fp_genes2))
n_fp <- length(fp_genes)
results <- append_results(results, "Arriba", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(fp_genes%in% bgenes_grx$genes)))[1] "Background genes in FPs: 0"plot_simu_benchmarking(results, "Arriba")
StringTie
# load data and extract transcripts marked as novel
stringtie_results <- read.delim(here("data/simu/results/StringTie/fullsimu.gtf"), comment="#", header=F)
stringtie_results <- stringtie_results[!stringtie_results$V9 %like% "ref_gene_id",]
stringtie_results <- stringtie_results[stringtie_results$V3 == "transcript",]
chrom <- sapply(stringtie_results$V1, convert_chrom, add_chr = FALSE)
stringtie_grx <- GRanges(seqnames = chrom,
ranges = IRanges(start = stringtie_results$V4,
end = stringtie_results$V5))
hits <- get_hits_oneloc(stringtie_grx, fus_grx[[1]], fus_grx[[2]], tsv_nsv_grx, bgenes_grx)
found_fus <- table(fus_truth[hits$fus_truth_hits,]$fusion_type)
found_tsv <- table(tsv_nsv_truth[hits$tsv_nsv_truth_hits,]$vartype)
# get false positives
fp_grx <- stringtie_grx[!1:length(hits$caller_hits) %in% hits$caller_hits]
fp_gene_hits <- all_gene_grx[subjectHits(findOverlaps(fp_grx, all_gene_grx))]
n_fp <- length(unique(fp_gene_hits$genes))
results <- append_results(results, "StringTie", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", sum(hits$bg_hits %in% hits$caller_hits)))[1] "Background genes in FPs: 0"plot_simu_benchmarking(results, "StringTie")
KisSplice
# regenerate gene ranges objects "chr" prefix
# (this is required as KisSplice"s bam was aligned
# to a genome with chr prefixes)
fus_grx <- get_granges(fus_truth, convert_chrom = TRUE)
tsv_nsv_grx <- get_granges(tsv_nsv_truth, convert_chrom = TRUE)
bgenes_grx <- get_granges(bg_gene_ref, convert_chrom = FALSE)
# extract hits im fusion and TSV/NSV gene regions
bam <- here("data/simu/results/KisSplice/results.bam")
fus1_hits <- get_kissplice_hits(bam, fus_grx[[1]])
fus2_hits <- get_kissplice_hits(bam, fus_grx[[2]])
tsv_nsv_hits <- get_kissplice_hits(bam, tsv_nsv_grx)
# remame row names to match locations from results
rownames(fus_truth) <- sapply(fus_truth$loc1, convert_chrom)
rownames(tsv_nsv_truth) <- sapply(tsv_nsv_truth$loc, convert_chrom)
# tally results
found_fus <- table(fus_truth[names(fus1_hits)[fus1_hits | fus2_hits],]$fusion_type)
found_tsv <- table(tsv_nsv_truth[names(tsv_nsv_hits[tsv_nsv_hits]),]$vartype)
# count number of FP hits in background genes
# first, get all reads that we counted as "hits"
param <- ScanBamParam(which = c(fus_grx[[1]], fus_grx[[2]], tsv_nsv_grx),
what = c("pos"))
all_hits <- scanBam(bam, param = param)
all_hits <- unique(unlist(all_hits))
# now get all reads in the results and count all
# reads that were not counted as hits
param <- ScanBamParam(what = "pos")
all_results <- unique(scanBam(bam, param = param)[[1]]$pos)
fps <- all_results[!all_results %in% all_hits]
n_fp <- length(fps)
# count the numner of FPs in background genes
param <- ScanBamParam(which = bgenes_grx, what = c("pos"))
bg_hits <- scanBam(bam, param = param)
bg_hits <- bg_hits[as.numeric(bg_hits) %in% fps]
results <- append_results(results, "KisSplice", found_fus, found_tsv, n_fp)
print(paste("Background genes in FPs:", length(bg_hits)))[1] "Background genes in FPs: 0"plot_simu_benchmarking(results, "KisSplice")
All results
MINTIE paper Figure 3 containing results from all benchmarked methods.
mr <- melt(results, id.vars = c("vartype", "class"), variable.name = "method")
mr$method <- factor(mr$method,
levels=c("MINTIE", "TAP", "Barnacle", "SQUID", "JAFFA", "Arriba", "KisSplice", "StringTie"))
plot_all_simu_benchmarks(mr)
sessionInfo()R version 3.5.1 (2018-07-02)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS 10.14.6
Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib
locale:
[1] en_AU.UTF-8/en_AU.UTF-8/en_AU.UTF-8/C/en_AU.UTF-8/en_AU.UTF-8
attached base packages:
[1] parallel stats4 stats graphics grDevices utils datasets
[8] methods base
other attached packages:
[1] biomaRt_2.38.0 workflowr_1.6.1 Rsamtools_1.34.1
[4] Biostrings_2.50.2 XVector_0.22.0 GenomicRanges_1.34.0
[7] GenomeInfoDb_1.18.2 IRanges_2.16.0 S4Vectors_0.20.1
[10] BiocGenerics_0.28.0 ggplot2_3.1.0 here_0.1
[13] dplyr_0.8.1 data.table_1.12.0
loaded via a namespace (and not attached):
[1] Rcpp_1.0.1 prettyunits_1.0.2 assertthat_0.2.1
[4] rprojroot_1.3-2 digest_0.6.18 R6_2.4.0
[7] plyr_1.8.4 backports_1.1.3 RSQLite_2.1.1
[10] evaluate_0.13 httr_1.4.0 pillar_1.3.1
[13] zlibbioc_1.28.0 rlang_0.4.2 progress_1.2.0
[16] lazyeval_0.2.2 blob_1.1.1 rmarkdown_1.12
[19] labeling_0.3 BiocParallel_1.14.2 stringr_1.4.0
[22] RCurl_1.95-4.12 bit_1.1-14 munsell_0.5.0
[25] compiler_3.5.1 httpuv_1.5.2 xfun_0.5
[28] pkgconfig_2.0.2 htmltools_0.3.6 tidyselect_0.2.5
[31] tibble_2.1.1 GenomeInfoDbData_1.2.0 codetools_0.2-16
[34] XML_3.98-1.19 crayon_1.3.4 withr_2.1.2
[37] later_1.0.0 bitops_1.0-6 grid_3.5.1
[40] gtable_0.3.0 DBI_1.0.0 git2r_0.26.1
[43] magrittr_1.5 scales_1.0.0 stringi_1.4.3
[46] reshape2_1.4.3 fs_1.2.7 promises_1.1.0
[49] tools_3.5.1 bit64_0.9-7 Biobase_2.42.0
[52] glue_1.3.1 purrr_0.3.2 hms_0.4.2
[55] yaml_2.2.0 AnnotationDbi_1.44.0 colorspace_1.4-1
[58] memoise_1.1.0 knitr_1.22