In steverozen/SynSigGen: Create Catalogs of Synthetic Mutational Spectra
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
Overview and rationale
Checking for interference between low-burden and high-burden signatures,
using MMR deficient tumours as the source of high-burden signatures.
We will need to run SignatureAnalyzer first on the 500 tumors with
neither SBS26 nor SBS44, and then on the "hyper-mutated tumors,
which contain SBS1, SBS5, SBS26 and and SBS44.
We create two directories for SignatureAnayzer:
one directory for the SBS1 and SBS5-only tumors, one for the
"hypermutated" tumors with SBS1, SBS5, SBS27, and SBS44.
We create one directory for SigProfiler, containing a synthetic
catlog containing both the non-hyper and hyper tumors.
Required libraries
library(ICAMS)
library(SynSigGen)
Get actual exposures for high-mutational burden signatures, SBS26 and SBS44
maybe.hyper.samples <-
setdiff(colnames(sa.all.real.exposures),
colnames(sa.no.hyper.real.exposures))
maybe.hyper.colorectal <-
grep("ColoRect-AdenoCA::",maybe.hyper.samples,
fixed = TRUE, value = TRUE)
maybe.hyper.uterus <-
grep("Uterus-AdenoCA::",maybe.hyper.samples,
fixed = TRUE, value = TRUE)
maybe.hyper <- sp.all.real.exposures[ , c(maybe.hyper.colorectal, maybe.hyper.uterus)]
maybe.hyper2 <- maybe.hyper[ , maybe.hyper["SBS10a", ] < 10]
# SBS14 is Combined POLE and MMR deficient, so we do not use it.
sp.hyper.real.exp <- maybe.hyper2[ , maybe.hyper2["SBS14", ] < 10]
# Look for SBS44, SBS26; SBS6 is rare
sa.hyper.real.exp <- sa.all.real.exposures[ , colnames(sp.hyper.real.exp)]
sp.hyper.real.exp <- sp.hyper.real.exp[c("SBS26", "SBS44"), ]
sa.hyper.real.exp <-
sa.hyper.real.exp[c("BI_COMPOSITE_SBS26_S", "BI_COMPOSITE_SBS44_S"), ]
Compute parameters and create synthetic exposures for high-burden signatures
\code{SAAndSPSynDataOneCAType} will always include at least 1 non-zero exposure
in the the synthetic exposure output.
We put do this analysis in \code{../../syn.hyper.START},
and we will \strong{not} use the information in this directory later.
START.dir.root <- "syn.hyper.low.START"
SetNewOutDir(paste0("tmp.", START.dir.root),
overwrite = TRUE # For testing and development
)
set.seed(191908)
num.syn.tumors <- 500
hyper.info <-
SAAndSPSynDataOneCAType(
sa.hyper.real.exp,
sp.hyper.real.exp,
ca.type = NULL, # Do not filter by cancer type
num.syn.tumors,
file.prefix = "HYPER")
Double the exposures of the hyper tumors.
hyper.info$sa.syn.exp <- hyper.info$sa.syn.exp * 2
hyper.info$sp.syn.exp <- hyper.info$sp.syn.exp * 2
Get actual exposures for low-mutational burden signatures, SBS1 and SBS2
no.hyper.colorectal <-
grep("ColoRect-AdenoCA::", colnames(sa.no.hyper.real.exposures),
fixed = TRUE, value = TRUE)
sp.no.hyper.colorectal <- sp.all.real.exposures[ , no.hyper.colorectal]
mean(sp.no.hyper.colorectal["SBS1", ])
mean(sp.no.hyper.colorectal["SBS5", ])
no.hyper.uterus <-
grep("Uterus-AdenoCA::", colnames(sa.no.hyper.real.exposures),
fixed = TRUE, value = TRUE)
sp.no.hyper.uterus <- sp.all.real.exposures[ , no.hyper.uterus]
mean(sp.no.hyper.uterus["SBS1", ])
mean(sp.no.hyper.uterus["SBS5", ])
no.hyper.colnames <- c(no.hyper.colorectal, no.hyper.uterus)
sp.no.hyper.real.exp <-
sp.all.real.exposures[c("SBS1", "SBS5"), no.hyper.colnames]
stopifnot(dim(sp.no.hyper.real.exp) == c(2, 81)) # sanity check
sa.no.hyper.real.exp <-
sa.all.real.exposures[
c("BI_COMPOSITE_SBS1_P", "BI_COMPOSITE_SBS5_P"),
no.hyper.colnames]
There is no BI_COMPOSITE_SBS5_P in the SignatureAnalyzer exposures,
so we take the exposures from SBS5
sa.no.hyper.real.exp["BI_COMPOSITE_SBS5_P", ] <- sp.no.hyper.real.exp["SBS5", ]
# Alternative approach, not used
if (FALSE) {
sa.no.hyper.real.exp2 <- sa.no.hyper.real.exp[rowSums(sa.no.hyper.real.exp) > 0, ]
sp.sa.map.info <-
MapSPToSASignatureNamesInExposure(
sp.exposures = sp.no.hyper.real.exp,
sa.sig.names.to.consider = rownames(sa.no.hyper.real.exp2))
knitr::kable(
sp.sa.map.info$sp.to.sa.sig.match,
caption =
'Best matches from SP signatures to SA signatures with exposures in hyper tumors',
digits = 4)
sa.no.hyper.real.exp3 <- sa.no.hyper.real.exp2[sp.sa.map.info$sp.to.sa.sig.match$to, ]
}
# View(sa.all.real.exposures[ , no.hyper.colnames])
stopifnot(dim(sa.no.hyper.real.exp) == c(2, 81)) # sanity check
Compute parameters and create synthetic exposures for low-burden signatures
num.syn.tumors <- 500
no.hyper.syn.info <-
SAAndSPSynDataOneCAType(
sa.no.hyper.real.exp,
sp.no.hyper.real.exp,
ca.type = NULL, # Do not filter by cancer type
num.syn.tumors,
file.prefix = "NO.HYPER")
Exposures for a set of ~500 tumours with SBS1, SBS5, SBS26, and SBS44
Most tumors have a mix of all 4 signatures. This reflects
the SignatureAnalyzer strategy of dealing with the
"hypermutated" tumours separately.
Combine the high and low burden synthetic exposures.
sa.hyper.syn.exp <-
rbind(hyper.info$sa.syn.exp,
no.hyper.syn.info$sa.syn.exp)
sp.hyper.syn.exp <-
rbind(hyper.info$sp.syn.exp,
no.hyper.syn.info$sp.syn.exp)
Remove hyper-mutated exposures that contain neither SBS26 nor SBS44
sa.to.keep <-
(sa.hyper.syn.exp["BI_COMPOSITE_SBS26_S", ] >= 1) |
(sa.hyper.syn.exp["BI_COMPOSITE_SBS44_S", ] >= 1)
sa.hyper.syn.exp <- sa.hyper.syn.exp[ , sa.to.keep]
sp.to.keep <-
(sp.hyper.syn.exp["SBS26", ] >= 1 ) |
(sp.hyper.syn.exp["SBS44", ] >= 1)
sp.hyper.syn.exp <- sp.hyper.syn.exp[ , sp.to.keep]
Data set for SA -- one (SBS1, 5) catalog, one (SBS1, 5, 26, 44) catalog
Catalogs for 500 tumors with only SBS1, 5
Create catalogs based on SignatureAnalyzer attributions
sa.primary.dir.root <- "syn.hyper.low.sa.primary"
SetNewOutDir(paste0("tmp.", sa.primary.dir.root),
overwrite = TRUE # For testing and development
)
overwrite = TRUE
CreateAndWriteCatalog(
sa.COMPOSITE.sigs,
no.hyper.syn.info$sa.syn.exp,
"sa.sa.COMPOSITE",
WriteCatCOMPOSITE,
overwrite = overwrite)
CreateAndWriteCatalog(
sa.96.sigs,
no.hyper.syn.info$sa.syn.exp,
"sa.sa.96",
WriteCatalog,
overwrite = overwrite)
Create synthetic catalogs based on SigProfiler attributions
sp.no.hyper.exp.renamed <- no.hyper.syn.info$sp.syn.exp
rownames(sp.no.hyper.exp.renamed) <- rownames(no.hyper.syn.info$sa.syn.exp)
CreateAndWriteCatalog(
sa.COMPOSITE.sigs,
sp.no.hyper.exp.renamed,
"sp.sa.COMPOSITE",
WriteCatCOMPOSITE,
overwrite = overwrite)
CreateAndWriteCatalog(
sp.sigs,
no.hyper.syn.info$sp.syn.exp,
"sp.sp",
WriteCatalog,
overwrite = overwrite)
Catalogs for 500 tumors with SBS1, 5, 26, 44
Create catalogs based on SignatureAnalyzer attributions
sa.secondary.dir.root <- "syn.hyper.low.sa.secondary"
SetNewOutDir(paste0("tmp.", sa.secondary.dir.root),
overwrite = TRUE # For testing and development
)
CreateAndWriteCatalog(
sa.COMPOSITE.sigs,
sa.hyper.syn.exp,
"sa.sa.COMPOSITE",
WriteCatCOMPOSITE,
overwrite = overwrite)
CreateAndWriteCatalog(
sa.96.sigs,
sa.hyper.syn.exp,
"sa.sa.96",
WriteCatalog,
overwrite = overwrite)
Create synthetic catalogs based on SigProfiler attributions
sp.hyper.exp.renamed <- sp.hyper.syn.exp
rownames(sp.hyper.exp.renamed) <- rownames(sa.hyper.syn.exp)
CreateAndWriteCatalog(
sa.COMPOSITE.sigs,
sp.hyper.exp.renamed,
"sp.sa.COMPOSITE",
WriteCatCOMPOSITE,
overwrite = overwrite)
CreateAndWriteCatalog(
sp.sigs,
sp.hyper.syn.exp,
"sp.sp",
WriteCatalog,
overwrite = overwrite)
Generate data for SigProfiler
Write the catalogs for SP
sp.dir.root <- "syn.hyper.low.sp"
SetNewOutDir(paste0("tmp.", sp.dir.root),
overwrite = TRUE # For testing and development
)
sa.merge.syn.exp <-
MergeExposures(
list(no.hyper.syn.info$sa.syn.exp, sa.hyper.syn.exp))
sa.tmp <-
CreateAndWriteCatalog(
sa.96.sigs,
sa.merge.syn.exp,
"sa.sa.96",
WriteCatalog,
overwrite = overwrite)
sp.merge.syn.exp <-
MergeExposures(
list(no.hyper.syn.info$sp.syn.exp, sp.hyper.syn.exp))
sp.tmp <-
CreateAndWriteCatalog(
sp.sigs,
sp.merge.syn.exp,
"sp.sp",
WriteCatalog,
overwrite = overwrite)
Check and unlink the results
Set unlink = FALSE if you want to keep the results.
Return values are "ok" on success.
Diff4SynDataSets(START.dir.root, unlink = TRUE)
Diff4SynDataSets(sa.primary.dir.root, unlink = TRUE)
Diff4SynDataSets(sa.secondary.dir.root, unlink = TRUE)
Diff4SynDataSets(sp.dir.root, unlink = TRUE)
steverozen/SynSigGen documentation built on April 1, 2022, 8:54 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
knitr::opts_chunk$set( collapse = TRUE, comment = "#>" )
Overview and rationale
Checking for interference between low-burden and high-burden signatures, using MMR deficient tumours as the source of high-burden signatures.
We will need to run SignatureAnalyzer first on the 500 tumors with neither SBS26 nor SBS44, and then on the "hyper-mutated tumors, which contain SBS1, SBS5, SBS26 and and SBS44.
We create two directories for SignatureAnayzer: one directory for the SBS1 and SBS5-only tumors, one for the "hypermutated" tumors with SBS1, SBS5, SBS27, and SBS44.
We create one directory for SigProfiler, containing a synthetic catlog containing both the non-hyper and hyper tumors.
Required libraries
library(ICAMS) library(SynSigGen)
Get actual exposures for high-mutational burden signatures, SBS26 and SBS44
maybe.hyper.samples <- setdiff(colnames(sa.all.real.exposures), colnames(sa.no.hyper.real.exposures)) maybe.hyper.colorectal <- grep("ColoRect-AdenoCA::",maybe.hyper.samples, fixed = TRUE, value = TRUE) maybe.hyper.uterus <- grep("Uterus-AdenoCA::",maybe.hyper.samples, fixed = TRUE, value = TRUE) maybe.hyper <- sp.all.real.exposures[ , c(maybe.hyper.colorectal, maybe.hyper.uterus)] maybe.hyper2 <- maybe.hyper[ , maybe.hyper["SBS10a", ] < 10] # SBS14 is Combined POLE and MMR deficient, so we do not use it. sp.hyper.real.exp <- maybe.hyper2[ , maybe.hyper2["SBS14", ] < 10] # Look for SBS44, SBS26; SBS6 is rare sa.hyper.real.exp <- sa.all.real.exposures[ , colnames(sp.hyper.real.exp)] sp.hyper.real.exp <- sp.hyper.real.exp[c("SBS26", "SBS44"), ] sa.hyper.real.exp <- sa.hyper.real.exp[c("BI_COMPOSITE_SBS26_S", "BI_COMPOSITE_SBS44_S"), ]
Compute parameters and create synthetic exposures for high-burden signatures
\code{SAAndSPSynDataOneCAType} will always include at least 1 non-zero exposure in the the synthetic exposure output.
We put do this analysis in \code{../../syn.hyper.START}, and we will \strong{not} use the information in this directory later.
START.dir.root <- "syn.hyper.low.START" SetNewOutDir(paste0("tmp.", START.dir.root), overwrite = TRUE # For testing and development ) set.seed(191908) num.syn.tumors <- 500 hyper.info <- SAAndSPSynDataOneCAType( sa.hyper.real.exp, sp.hyper.real.exp, ca.type = NULL, # Do not filter by cancer type num.syn.tumors, file.prefix = "HYPER")
Double the exposures of the hyper tumors.
hyper.info$sa.syn.exp <- hyper.info$sa.syn.exp * 2 hyper.info$sp.syn.exp <- hyper.info$sp.syn.exp * 2
Get actual exposures for low-mutational burden signatures, SBS1 and SBS2
no.hyper.colorectal <- grep("ColoRect-AdenoCA::", colnames(sa.no.hyper.real.exposures), fixed = TRUE, value = TRUE) sp.no.hyper.colorectal <- sp.all.real.exposures[ , no.hyper.colorectal] mean(sp.no.hyper.colorectal["SBS1", ]) mean(sp.no.hyper.colorectal["SBS5", ]) no.hyper.uterus <- grep("Uterus-AdenoCA::", colnames(sa.no.hyper.real.exposures), fixed = TRUE, value = TRUE) sp.no.hyper.uterus <- sp.all.real.exposures[ , no.hyper.uterus] mean(sp.no.hyper.uterus["SBS1", ]) mean(sp.no.hyper.uterus["SBS5", ]) no.hyper.colnames <- c(no.hyper.colorectal, no.hyper.uterus) sp.no.hyper.real.exp <- sp.all.real.exposures[c("SBS1", "SBS5"), no.hyper.colnames] stopifnot(dim(sp.no.hyper.real.exp) == c(2, 81)) # sanity check sa.no.hyper.real.exp <- sa.all.real.exposures[ c("BI_COMPOSITE_SBS1_P", "BI_COMPOSITE_SBS5_P"), no.hyper.colnames]
There is no BI_COMPOSITE_SBS5_P in the SignatureAnalyzer exposures, so we take the exposures from SBS5
sa.no.hyper.real.exp["BI_COMPOSITE_SBS5_P", ] <- sp.no.hyper.real.exp["SBS5", ]
# Alternative approach, not used if (FALSE) { sa.no.hyper.real.exp2 <- sa.no.hyper.real.exp[rowSums(sa.no.hyper.real.exp) > 0, ] sp.sa.map.info <- MapSPToSASignatureNamesInExposure( sp.exposures = sp.no.hyper.real.exp, sa.sig.names.to.consider = rownames(sa.no.hyper.real.exp2)) knitr::kable( sp.sa.map.info$sp.to.sa.sig.match, caption = 'Best matches from SP signatures to SA signatures with exposures in hyper tumors', digits = 4) sa.no.hyper.real.exp3 <- sa.no.hyper.real.exp2[sp.sa.map.info$sp.to.sa.sig.match$to, ] }
# View(sa.all.real.exposures[ , no.hyper.colnames]) stopifnot(dim(sa.no.hyper.real.exp) == c(2, 81)) # sanity check
Compute parameters and create synthetic exposures for low-burden signatures
num.syn.tumors <- 500 no.hyper.syn.info <- SAAndSPSynDataOneCAType( sa.no.hyper.real.exp, sp.no.hyper.real.exp, ca.type = NULL, # Do not filter by cancer type num.syn.tumors, file.prefix = "NO.HYPER")
Exposures for a set of ~500 tumours with SBS1, SBS5, SBS26, and SBS44
Most tumors have a mix of all 4 signatures. This reflects the SignatureAnalyzer strategy of dealing with the "hypermutated" tumours separately.
Combine the high and low burden synthetic exposures.
sa.hyper.syn.exp <- rbind(hyper.info$sa.syn.exp, no.hyper.syn.info$sa.syn.exp) sp.hyper.syn.exp <- rbind(hyper.info$sp.syn.exp, no.hyper.syn.info$sp.syn.exp)
Remove hyper-mutated exposures that contain neither SBS26 nor SBS44
sa.to.keep <- (sa.hyper.syn.exp["BI_COMPOSITE_SBS26_S", ] >= 1) | (sa.hyper.syn.exp["BI_COMPOSITE_SBS44_S", ] >= 1) sa.hyper.syn.exp <- sa.hyper.syn.exp[ , sa.to.keep] sp.to.keep <- (sp.hyper.syn.exp["SBS26", ] >= 1 ) | (sp.hyper.syn.exp["SBS44", ] >= 1) sp.hyper.syn.exp <- sp.hyper.syn.exp[ , sp.to.keep]
Data set for SA -- one (SBS1, 5) catalog, one (SBS1, 5, 26, 44) catalog
Catalogs for 500 tumors with only SBS1, 5
Create catalogs based on SignatureAnalyzer attributions
sa.primary.dir.root <- "syn.hyper.low.sa.primary" SetNewOutDir(paste0("tmp.", sa.primary.dir.root), overwrite = TRUE # For testing and development ) overwrite = TRUE CreateAndWriteCatalog( sa.COMPOSITE.sigs, no.hyper.syn.info$sa.syn.exp, "sa.sa.COMPOSITE", WriteCatCOMPOSITE, overwrite = overwrite) CreateAndWriteCatalog( sa.96.sigs, no.hyper.syn.info$sa.syn.exp, "sa.sa.96", WriteCatalog, overwrite = overwrite)
Create synthetic catalogs based on SigProfiler attributions
sp.no.hyper.exp.renamed <- no.hyper.syn.info$sp.syn.exp rownames(sp.no.hyper.exp.renamed) <- rownames(no.hyper.syn.info$sa.syn.exp) CreateAndWriteCatalog( sa.COMPOSITE.sigs, sp.no.hyper.exp.renamed, "sp.sa.COMPOSITE", WriteCatCOMPOSITE, overwrite = overwrite)
CreateAndWriteCatalog( sp.sigs, no.hyper.syn.info$sp.syn.exp, "sp.sp", WriteCatalog, overwrite = overwrite)
Catalogs for 500 tumors with SBS1, 5, 26, 44
Create catalogs based on SignatureAnalyzer attributions
sa.secondary.dir.root <- "syn.hyper.low.sa.secondary" SetNewOutDir(paste0("tmp.", sa.secondary.dir.root), overwrite = TRUE # For testing and development ) CreateAndWriteCatalog( sa.COMPOSITE.sigs, sa.hyper.syn.exp, "sa.sa.COMPOSITE", WriteCatCOMPOSITE, overwrite = overwrite) CreateAndWriteCatalog( sa.96.sigs, sa.hyper.syn.exp, "sa.sa.96", WriteCatalog, overwrite = overwrite)
Create synthetic catalogs based on SigProfiler attributions
sp.hyper.exp.renamed <- sp.hyper.syn.exp rownames(sp.hyper.exp.renamed) <- rownames(sa.hyper.syn.exp) CreateAndWriteCatalog( sa.COMPOSITE.sigs, sp.hyper.exp.renamed, "sp.sa.COMPOSITE", WriteCatCOMPOSITE, overwrite = overwrite)
CreateAndWriteCatalog( sp.sigs, sp.hyper.syn.exp, "sp.sp", WriteCatalog, overwrite = overwrite)
Generate data for SigProfiler
Write the catalogs for SP
sp.dir.root <- "syn.hyper.low.sp" SetNewOutDir(paste0("tmp.", sp.dir.root), overwrite = TRUE # For testing and development ) sa.merge.syn.exp <- MergeExposures( list(no.hyper.syn.info$sa.syn.exp, sa.hyper.syn.exp)) sa.tmp <- CreateAndWriteCatalog( sa.96.sigs, sa.merge.syn.exp, "sa.sa.96", WriteCatalog, overwrite = overwrite)
sp.merge.syn.exp <- MergeExposures( list(no.hyper.syn.info$sp.syn.exp, sp.hyper.syn.exp)) sp.tmp <- CreateAndWriteCatalog( sp.sigs, sp.merge.syn.exp, "sp.sp", WriteCatalog, overwrite = overwrite)
Check and unlink the results
Set unlink = FALSE if you want to keep the results.
Return values are "ok" on success.
Diff4SynDataSets(START.dir.root, unlink = TRUE) Diff4SynDataSets(sa.primary.dir.root, unlink = TRUE) Diff4SynDataSets(sa.secondary.dir.root, unlink = TRUE) Diff4SynDataSets(sp.dir.root, unlink = TRUE)
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.