In ProMetIS/ProMetIA: Proteomics and Metabolomics Integrated Analysis
knitr::opts_chunk$set(fig.width = 8,
fig.height = 8,
fig.path = 'figures/temp/1_post_processing/',
message = FALSE,
warning = FALSE)
Mice IDs
mice_id.df <- read.table(file.path(LATMX2::processed_dir.c(),
"mice_id.tsv"),
row.names = 1,
header = TRUE, sep = "\t")
mice_id.vc <- rownames(mice_id.df)
mice_num.vc <- substr(mice_id.vc, 2, 4)
stopifnot(identical(as.integer(mice_num.vc),
sort(as.integer(mice_num.vc))))
Clinics
Loading
clinic_setname.c <- "clinics"
# table from PHENOMIN (47 * 1098)
clinic_raw.df <- read.table(file.path(LATMX2::processed_dir.c(),
clinic_setname.c,
"phenomin.tsv"),
check.names = FALSE, comment.char = "",
header = TRUE, quote = "", sep = "\t")
clinic_proc_names.vc <- colnames(clinic_raw.df)
colnames(clinic_raw.df) <- make.names(colnames(clinic_raw.df), unique = TRUE)
# checking sample order
stopifnot(identical(clinic_raw.df[, "mouse_nb"],
sort(clinic_raw.df[, "mouse_nb"])))
sampleMetadata
# restricting to the 42 samples analyzed in ProMetIS
clinic_raw.df <- clinic_raw.df[as.character(clinic_raw.df[, "mouse_nb"]) %in% mice_num.vc, ]
stopifnot(identical(as.character(clinic_raw.df[, "mouse_nb"]), mice_num.vc))
rownames(clinic_raw.df) <- mice_id.vc
# sampleMetadata (42 * 1098)
clinic_pda.df <- clinic_raw.df
# labeling the 'supplementary' information (to be kept at the end of the
# sampleMetadata and variableMetadata tables)
supp.vi <- which(!(colnames(clinic_pda.df) %in% colnames(mice_id.df)))
stopifnot(length(supp.vi) == 1094)
colnames(clinic_pda.df)[supp.vi] <- paste0("supp_",
colnames(clinic_pda.df)[supp.vi])
dataMatrix
# initial dataMatrix (42 * 1091)
variable.vi <- which(!(colnames(clinic_raw.df) %in% c(colnames(mice_id.df),
c("mouse_id",
"genotype",
"project"))))
stopifnot(length(variable.vi) == 1091)
clinic_dat.df <- clinic_raw.df[, variable.vi]
dat_proc_names.vc <- clinic_proc_names.vc[variable.vi]
rm(clinic_raw.df)
rm(clinic_proc_names.vc)
## discarding variables which cannot be converted to numerics
### removing 1 variable with all values identical except 1 (for "W627m")
### note that all 'Test.de.Tolérance.au.Glucose' tests have a 'NA' value for this mouse only
discard_imagerie.vl <- grepl("Imagerie.du.squelette.par.rayons.X.Digits.integrity",
colnames(clinic_dat.df))
stopifnot(sum(discard_imagerie.vl) == 1)
### removing date of sacrifice
discard_sacrifice.vl <- grepl("Date.of.sacrifice", colnames(clinic_dat.df))
stopifnot(sum(discard_sacrifice.vl) == 2)
### removing 471 columns with NA or "" values only
discard_navoid.vl <- apply(clinic_dat.df, 2,
function(y) {
unique.vc <- unique(as.character(y))
unique.vc <- unique.vc[!is.na(unique.vc) & (unique.vc != "")]
length(unique.vc) < 2
})
stopifnot(sum(discard_navoid.vl) == 478)
discard_feat.vl <- discard_imagerie.vl |
discard_sacrifice.vl |
discard_navoid.vl
stopifnot(sum(discard_feat.vl) == 481)
clinic_dat.df <- clinic_dat.df[, !discard_feat.vl]
dat_proc_names.vc <- dat_proc_names.vc[!discard_feat.vl]
## keeping numerics and factors with nlevels > 2
clinic_class.vc <- sapply(clinic_dat.df, data.class)
stopifnot(identical(unique(clinic_class.vc), c("numeric", "factor")))
## selecting numerics (678)
keep_numeric.vl <- clinic_class.vc == "numeric"
stopifnot(sum(keep_numeric.vl) == 571)
## selecting factors with 2 levels (45)
keep_factor2.vl <- logical(ncol(clinic_dat.df))
for (j in which(clinic_class.vc == "factor")) {
clinic_dat.f <- clinic_dat.df[, j]
if (nlevels(clinic_dat.f) == 2) {
clinic_dat.df[, j] <- as.integer(clinic_dat.df[, j])
keep_factor2.vl[j] <- TRUE
}
}
stopifnot(sum(keep_factor2.vl) == 1)
## aggregating selections (725)
keep_aggreg.vl <- keep_numeric.vl |
keep_factor2.vl
stopifnot(sum(keep_aggreg.vl) == 572)
clinic_dat.mn <- as.matrix(clinic_dat.df[, keep_aggreg.vl])
dat_proc_names.vc <- dat_proc_names.vc[keep_aggreg.vl]
## removing variables with > 20% NA (including 100% NA in females)
## or with variance < 1e-5
keep_notna_zerovar.vl <- ProMetIA:::.filter_na_zerovar(clinic_dat.mn,
na_thresh.n = 0.2,
var_thresh.n = 1e-5)
stopifnot(sum(keep_notna_zerovar.vl) == 213)
clinic_dat.mn <- clinic_dat.mn[, keep_notna_zerovar.vl]
dat_proc_names.vc <- dat_proc_names.vc[keep_notna_zerovar.vl]
stopifnot(identical(dim(clinic_dat.mn), as.integer(c(42, 213))))
variableMetadata
# variableMetadata (238 * 2)
category.vc <- sapply(colnames(clinic_dat.mn),
function(name.c)
unlist(strsplit(name.c,
split = ".",
fixed = TRUE))[1])
category.vc <- gsub("Auditory", "Auditory and PPI",
gsub("Clinical", "Clinical chemistry",
gsub("Grip", "Grip test",
gsub("Hot", "Hot plate",
gsub("Modified", "SHIRPA",
gsub("Open", "Open field",
gsub("Pavlovian", "Pavlovian fear cond.",
gsub("Shock", "Shock threshold",
gsub("Test", "Glucose tolerance test",
gsub("Ultrafocus", "Body composition",
category.vc))))))))))
full.vc <- sapply(colnames(clinic_dat.mn),
function(name.c) {
name.vc <- unlist(strsplit(name.c,
split = ".",
fixed = TRUE))
if (length(name.vc) == 1) {
return("")
} else {
name.vc <- name.vc[-1]
name.vc <- name.vc[name.vc != ""]
return(paste(name.vc, collapse = "_"))
}
})
clinic_fda.df <- data.frame(row.names = colnames(clinic_dat.mn),
initial_names = dat_proc_names.vc,
category = category.vc,
full_info = full.vc,
stringsAsFactors = FALSE)
stopifnot(identical(dim(clinic_fda.df), as.integer(c(213, 3))))
ExpressionSet
# converting to ExpressionSet
clinic.eset <- Biobase::ExpressionSet(assayData = t(clinic_dat.mn),
phenoData = Biobase::AnnotatedDataFrame(data = clinic_pda.df),
featureData = Biobase::AnnotatedDataFrame(data = clinic_fda.df),
experimentData = Biobase::MIAME(title = clinic_setname.c))
stopifnot(methods::validObject(clinic.eset))
print(clinic.eset)
Saving (not run)
phenomis::writing(clinic.eset,
file.path("../../LATMX2/inst/extdata/2_post_processed",
clinic_setname.c),
overwrite.l = TRUE)
Proteomics
Set names and input files
proteo_files.vc <- vapply(LATMX2::proteo_sets.vc(),
function(set.c) {
file.c <- list.files(file.path(LATMX2::processed_dir.c(), set.c),
pattern = ".xlsx", full.names = TRUE)
stopifnot(length(file.c) == 1)
file.c
}, FUN.VALUE = character(1))
proteo.mset <- MultiDataSet::createMultiDataSet()
MultiDataSet containing the data matrices
for (set.c in LATMX2::proteo_sets.vc()) {
# dataMatrix
data.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c],
sheet = 1),
stringsAsFactors = FALSE)
rownames(data.df) <- data.df[, 1]
data.df[, 1] <- NULL
data.mn <- as.matrix(data.df)
rm(data.df)
mode(data.mn) <- "numeric"
eset <- Biobase::ExpressionSet(assayData = data.mn,
experimentData = Biobase::MIAME(title = set.c))
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
sampleMetadata
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
proteo_pda.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c],
sheet = 2),
stringsAsFactors = FALSE)
rownames(proteo_pda.df) <- gsub(".", "_",
proteo_pda.df[, 1], fixed = TRUE)
sample_names.vc <- Biobase::sampleNames(eset)
if (set.c == "proteomics_liver") {
stopifnot(identical(sort(rownames(proteo_pda.df)),
sort(sample_names.vc)))
proteo_pda.df <- proteo_pda.df[sample_names.vc, ]
}
stopifnot(identical(rownames(proteo_pda.df),
sample_names.vc))
colnames(proteo_pda.df) <- paste0("supp_", colnames(proteo_pda.df))
Biobase::pData(eset) <- proteo_pda.df
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
variableMetadata
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
proteo_fda.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c],
sheet = 3),
stringsAsFactors = FALSE)
rownames(proteo_fda.df) <- proteo_fda.df[, 1]
stopifnot(identical(rownames(proteo_fda.df),
Biobase::featureNames(eset)))
proteo_fda.df[, "uniprot_id"] <- sapply(proteo_fda.df[, "accession"],
function(access.c)
unlist(strsplit(access.c,
split = "|",
fixed = TRUE))[2],
USE.NAMES = FALSE)
supp.vi <- which(!(colnames(proteo_fda.df) %in% c("accession",
"description",
"uniprot_id")))
colnames(proteo_fda.df)[supp.vi] <- paste0("supp_",
colnames(proteo_fda.df)[supp.vi])
Biobase::fData(eset) <- proteo_fda.df
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
Renaming features and re-ordering samples
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
# changing variable IDs to names
feat_names.vc <- paste0(sapply(Biobase::fData(eset)[, "accession"],
function(access.c)
unlist(strsplit(access.c, split = "|",
fixed = TRUE))[2]),
"_",
sapply(Biobase::fData(eset)[, "description"],
function(desc.c) {
if (!is.na(desc.c) && desc.c != "") {
desc.c <- unlist(strsplit(desc.c, split = " OS="))[1]
if (nchar(desc.c) > 25)
desc.c <- paste0(substr(desc.c, 1, 24), ".")
return(desc.c)
}
}))
stopifnot(!any(duplicated(feat_names.vc)))
Biobase::featureNames(eset) <- feat_names.vc
# discarding pools
if (set.c == "proteomics_liver")
eset <- eset[, !grepl("(p|P)ool", Biobase::pData(eset)[, "supp_sample name"])]
# re-ordering samples
if (set.c == "proteomics_liver") {
eset <- eset[, order(as.integer(Biobase::pData(eset)[, "supp_sample name"]))]
} else {
Biobase::sampleNames(eset) <- sapply(Biobase::sampleNames(eset),
function(samp.c) unlist(strsplit(samp.c, split = "_"))[2], USE.NAMES = FALSE)
eset <- eset[, order(as.integer(Biobase::sampleNames(eset)))]
}
# renaming samples
if (set.c == "proteomics_liver") {
Biobase::sampleNames(eset) <- paste0(sapply(Biobase::pData(eset)[, "supp_Condition"],
function(cond.c)
switch(cond.c,
mx2 = "X",
ctrl = "W",
lat = "L")),
Biobase::pData(eset)[, "supp_sample name"])
} else {
Biobase::sampleNames(eset) <- paste0(sapply(Biobase::pData(eset)[, "supp_Condition"],
function(cond.c)
switch(cond.c,
Mx2 = "X",
CONT = "W",
LAT = "L")),
Biobase::sampleNames(eset))
}
mice_konum.vc <- substr(mice_id.vc, 1, 4)
mice_koid.vc <- mice_id.vc
if (set.c == "proteomics_plasma") {
stopifnot(all(Biobase::sampleNames(eset) %in% mice_konum.vc))
mice_koid.vc <- mice_koid.vc[mice_konum.vc %in% Biobase::sampleNames(eset)]
mice_konum.vc <- mice_konum.vc[mice_konum.vc %in% Biobase::sampleNames(eset)]
}
stopifnot(identical(Biobase::sampleNames(eset),
mice_konum.vc))
Biobase::sampleNames(eset) <- mice_koid.vc
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
Adding meta data (gene, mouse_nb, sex)
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
Biobase::pData(eset) <- cbind.data.frame(mice_id.df[Biobase::sampleNames(eset), ],
Biobase::pData(eset),
stringsAsFactors = FALSE)
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
formatting the 'imputation' metrics
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
eset <- ProMetIA::format_imputation(eset)
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
Variable metadata: including new mice ids in the column names (count, abundance, etc.)
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
samp.vc <- Biobase::sampleNames(eset)
pdata.df <- Biobase::pData(eset)
fvar.vc <- Biobase::fvarLabels(eset)
for (i in seq_along(samp.vc)) {
samp.c <- samp.vc[i]
samp_id.c <- gsub("abundance_", "", pdata.df[i, "id"])
fvar_id.vi <- grep(paste0("^supp_.+", samp_id.c), fvar.vc)
fvar.vc[fvar_id.vi] <- gsub(switch(set.c,
proteomics_liver = "mgf",
proteomics_plasma = samp_id.c),
samp.c, fvar.vc[fvar_id.vi])
}
Biobase::fvarLabels(eset) <- fvar.vc
stopifnot(methods::validObject(eset))
proteo.mset <- MultiDataSet::add_eset(proteo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
Re-ordering metadata ('supp_' columns at the end)
proteo.mset <- ProMetIA::ordermeta_mset(proteo.mset)
Saving (not run)
for (set.c in LATMX2::proteo_sets.vc()) {
eset <- proteo.mset[[set.c]]
# an 'id' extra column has been automatically created in the sampleMetadata by
# 'MultiDataSet' and must be removed before saving
pdata.df <- Biobase::pData(eset)
pdata.df[, "id"] <- NULL
Biobase::pData(eset) <- pdata.df
phenomis::writing(eset,
file.path("../../LATMX2/inst/extdata/2_post_processed",
set.c),
overwrite.l = TRUE)
}
Metabolomics (in progress)
Loading
metabo.mset <- phenomis::reading(file.path(LATMX2::processed_dir.c()),
subsets.vc = LATMX2::metabo_sets.vc(),
report.c = "none")
Normalization of the dataMatrix
for (set.c in LATMX2::metabo_sets.vc()) {
eset <- metabo.mset[[set.c]]
if (grepl("(hyper|hilic)", set.c)) {
if (grepl("c18hyper", set.c)) {
# setting NA values to 0
exprs.mn <- Biobase::exprs(eset)
exprs.mn[is.na(exprs.mn)] <- ifelse(set.c == "metabolomics_liver_c18hyper-pos", 0, 1)
Biobase::exprs(eset) <- exprs.mn
}
# computing quality metrics
eset <- phenomis::inspecting(eset, title.c = set.c)
# discarding features with mean intensity in samples / mean intensity in pools < 3
eset <- eset[Biobase::fData(eset)[, "blankMean_over_sampleMean"] <= 0.33, ]
# discarding features with a Pearson correlation with pool dilution < 0.7
eset <- eset[Biobase::fData(eset)[, "poolDil_cor"] >= 0.7, ]
} else if (grepl("acqui", set.c)) {
# discarding features eluted before 0.4 min or after 22 min
eset <- eset[Biobase::fData(eset)[, "rt"] >= 0.4 & Biobase::fData(eset)[, "rt"] <= 22, ]
# discarding features with mean intensity in samples / mean intensity in pools < 2 (or 4)
fold.i <- ifelse(set.c == "metabolomics_plasma_c18acqui-pos", 2, 4)
eset <- eset[Biobase::fData(eset)[, "fold"] >= fold.i, ]
eset <- eset[Biobase::fData(eset)[, "tstat"] <= 0, ]
} else
stop("Unknown metabolomics dataset name.")
# discarding blanks and diluted pools (QCs)
eset <- eset[, Biobase::pData(eset)[, "sampleType"] %in% c("sample", "pool")]
# signal drift correction
if (set.c == "metabolomics_liver_c18hyper-pos") {
reference.c <- "none"
} else if (set.c == "metabolomics_plasma_c18hyper-pos") {
reference.c <- "pool"
} else if (set.c == "metabolomics_liver_hilic-neg") {
reference.c <- "none"
} else if (set.c == "metabolomics_plasma_hilic-neg") {
reference.c <- "none"
} else if (set.c == "metabolomics_plasma_c18acqui-pos") {
reference.c <- "sample"
} else if (set.c == "metabolomics_plasma_c18acqui-neg") {
reference.c <- "pool"
} else
stop()
if (reference.c != "none")
eset <- phenomis::correcting(eset, reference.c = reference.c, title.c = set.c)
# updating quality metrics
eset <- phenomis::inspecting(eset, title.c = set.c)
# discarding features with a coefficient of variation in pools > 30%
eset <- eset[Biobase::fData(eset)[, "pool_CV"] <= 0.3, ]
if (grepl("acqui", set.c)) {
# discarding features with a ratio 'coefficient of variation in samples / coefficient of variation in pools' < 0.8
eset <- eset[Biobase::fData(eset)[, "poolCV_over_sampleCV"] <= 1.25, ]
}
# discarding 'pool' samples (i.e. QCs)
eset <- eset[, Biobase::pData(eset)[, "sampleType"] != "pool"]
if (grepl("acqui", set.c)) {
# discarding the few features containing NA (generated after the drift correction)
eset <- eset[apply(Biobase::exprs(eset), 1, function(feat.vn) !any(is.na(feat.vn))), ]
# discarding features highly correlated to others from the same pcgroup with higher mean intensity
eset <- ProMetIA::filter_pcgroup(eset)
if (set.c == "metabolomics_plasma_c18acqui-pos") {
# discarding features with 90% quantile < 5000
eset <- eset[apply(Biobase::exprs(eset), 1,
function(feat.vn) quantile(feat.vn, 0.9)) >= 5000, ]
}
}
stopifnot(methods::validObject(eset))
metabo.mset <- MultiDataSet::add_eset(metabo.mset,
eset,
dataset.type = set.c,
GRanges = NA,
overwrite = TRUE,
warnings = FALSE)
}
Formatting and ordering sample names
metabo.mset <- ProMetIA::format_metabonames(metabo.mset,
mice_id.df = mice_id.df,
mice_id.vc = mice_id.vc,
mice_num.vc = mice_num.vc)
Log2 transformation
metabo.mset <- phenomis::transforming(metabo.mset)
Re-ordering metadata ('supp_' columns at the end)
metabo.mset <- ProMetIA::ordermeta_mset(metabo.mset)
Saving (not run)
for (set.c in LATMX2::metabo_sets.vc()) {
eset <- metabo.mset[[set.c]]
# an 'id' extra column has been automatically created in the sampleMetadata by
# 'MultiDataSet' and must be removed before saving
pdata.df <- Biobase::pData(eset)
pdata.df[, "id"] <- NULL
Biobase::pData(eset) <- pdata.df
phenomis::writing(eset,
file.path("../../LATMX2/inst/extdata/2_post_processed",
set.c),
overwrite.l = TRUE)
}
ProMetIS/ProMetIA documentation built on March 6, 2020, 2:11 a.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(fig.width = 8, fig.height = 8, fig.path = 'figures/temp/1_post_processing/', message = FALSE, warning = FALSE)
Mice IDs
mice_id.df <- read.table(file.path(LATMX2::processed_dir.c(), "mice_id.tsv"), row.names = 1, header = TRUE, sep = "\t") mice_id.vc <- rownames(mice_id.df) mice_num.vc <- substr(mice_id.vc, 2, 4) stopifnot(identical(as.integer(mice_num.vc), sort(as.integer(mice_num.vc))))
Clinics
Loading
clinic_setname.c <- "clinics" # table from PHENOMIN (47 * 1098) clinic_raw.df <- read.table(file.path(LATMX2::processed_dir.c(), clinic_setname.c, "phenomin.tsv"), check.names = FALSE, comment.char = "", header = TRUE, quote = "", sep = "\t") clinic_proc_names.vc <- colnames(clinic_raw.df) colnames(clinic_raw.df) <- make.names(colnames(clinic_raw.df), unique = TRUE) # checking sample order stopifnot(identical(clinic_raw.df[, "mouse_nb"], sort(clinic_raw.df[, "mouse_nb"])))
sampleMetadata
# restricting to the 42 samples analyzed in ProMetIS clinic_raw.df <- clinic_raw.df[as.character(clinic_raw.df[, "mouse_nb"]) %in% mice_num.vc, ] stopifnot(identical(as.character(clinic_raw.df[, "mouse_nb"]), mice_num.vc)) rownames(clinic_raw.df) <- mice_id.vc # sampleMetadata (42 * 1098) clinic_pda.df <- clinic_raw.df # labeling the 'supplementary' information (to be kept at the end of the # sampleMetadata and variableMetadata tables) supp.vi <- which(!(colnames(clinic_pda.df) %in% colnames(mice_id.df))) stopifnot(length(supp.vi) == 1094) colnames(clinic_pda.df)[supp.vi] <- paste0("supp_", colnames(clinic_pda.df)[supp.vi])
dataMatrix
# initial dataMatrix (42 * 1091) variable.vi <- which(!(colnames(clinic_raw.df) %in% c(colnames(mice_id.df), c("mouse_id", "genotype", "project")))) stopifnot(length(variable.vi) == 1091) clinic_dat.df <- clinic_raw.df[, variable.vi] dat_proc_names.vc <- clinic_proc_names.vc[variable.vi] rm(clinic_raw.df) rm(clinic_proc_names.vc) ## discarding variables which cannot be converted to numerics ### removing 1 variable with all values identical except 1 (for "W627m") ### note that all 'Test.de.Tolérance.au.Glucose' tests have a 'NA' value for this mouse only discard_imagerie.vl <- grepl("Imagerie.du.squelette.par.rayons.X.Digits.integrity", colnames(clinic_dat.df)) stopifnot(sum(discard_imagerie.vl) == 1) ### removing date of sacrifice discard_sacrifice.vl <- grepl("Date.of.sacrifice", colnames(clinic_dat.df)) stopifnot(sum(discard_sacrifice.vl) == 2) ### removing 471 columns with NA or "" values only discard_navoid.vl <- apply(clinic_dat.df, 2, function(y) { unique.vc <- unique(as.character(y)) unique.vc <- unique.vc[!is.na(unique.vc) & (unique.vc != "")] length(unique.vc) < 2 }) stopifnot(sum(discard_navoid.vl) == 478) discard_feat.vl <- discard_imagerie.vl | discard_sacrifice.vl | discard_navoid.vl stopifnot(sum(discard_feat.vl) == 481) clinic_dat.df <- clinic_dat.df[, !discard_feat.vl] dat_proc_names.vc <- dat_proc_names.vc[!discard_feat.vl] ## keeping numerics and factors with nlevels > 2 clinic_class.vc <- sapply(clinic_dat.df, data.class) stopifnot(identical(unique(clinic_class.vc), c("numeric", "factor"))) ## selecting numerics (678) keep_numeric.vl <- clinic_class.vc == "numeric" stopifnot(sum(keep_numeric.vl) == 571) ## selecting factors with 2 levels (45) keep_factor2.vl <- logical(ncol(clinic_dat.df)) for (j in which(clinic_class.vc == "factor")) { clinic_dat.f <- clinic_dat.df[, j] if (nlevels(clinic_dat.f) == 2) { clinic_dat.df[, j] <- as.integer(clinic_dat.df[, j]) keep_factor2.vl[j] <- TRUE } } stopifnot(sum(keep_factor2.vl) == 1) ## aggregating selections (725) keep_aggreg.vl <- keep_numeric.vl | keep_factor2.vl stopifnot(sum(keep_aggreg.vl) == 572) clinic_dat.mn <- as.matrix(clinic_dat.df[, keep_aggreg.vl]) dat_proc_names.vc <- dat_proc_names.vc[keep_aggreg.vl] ## removing variables with > 20% NA (including 100% NA in females) ## or with variance < 1e-5 keep_notna_zerovar.vl <- ProMetIA:::.filter_na_zerovar(clinic_dat.mn, na_thresh.n = 0.2, var_thresh.n = 1e-5) stopifnot(sum(keep_notna_zerovar.vl) == 213) clinic_dat.mn <- clinic_dat.mn[, keep_notna_zerovar.vl] dat_proc_names.vc <- dat_proc_names.vc[keep_notna_zerovar.vl] stopifnot(identical(dim(clinic_dat.mn), as.integer(c(42, 213))))
variableMetadata
# variableMetadata (238 * 2) category.vc <- sapply(colnames(clinic_dat.mn), function(name.c) unlist(strsplit(name.c, split = ".", fixed = TRUE))[1]) category.vc <- gsub("Auditory", "Auditory and PPI", gsub("Clinical", "Clinical chemistry", gsub("Grip", "Grip test", gsub("Hot", "Hot plate", gsub("Modified", "SHIRPA", gsub("Open", "Open field", gsub("Pavlovian", "Pavlovian fear cond.", gsub("Shock", "Shock threshold", gsub("Test", "Glucose tolerance test", gsub("Ultrafocus", "Body composition", category.vc)))))))))) full.vc <- sapply(colnames(clinic_dat.mn), function(name.c) { name.vc <- unlist(strsplit(name.c, split = ".", fixed = TRUE)) if (length(name.vc) == 1) { return("") } else { name.vc <- name.vc[-1] name.vc <- name.vc[name.vc != ""] return(paste(name.vc, collapse = "_")) } }) clinic_fda.df <- data.frame(row.names = colnames(clinic_dat.mn), initial_names = dat_proc_names.vc, category = category.vc, full_info = full.vc, stringsAsFactors = FALSE) stopifnot(identical(dim(clinic_fda.df), as.integer(c(213, 3))))
ExpressionSet
# converting to ExpressionSet clinic.eset <- Biobase::ExpressionSet(assayData = t(clinic_dat.mn), phenoData = Biobase::AnnotatedDataFrame(data = clinic_pda.df), featureData = Biobase::AnnotatedDataFrame(data = clinic_fda.df), experimentData = Biobase::MIAME(title = clinic_setname.c)) stopifnot(methods::validObject(clinic.eset)) print(clinic.eset)
Saving (not run)
phenomis::writing(clinic.eset, file.path("../../LATMX2/inst/extdata/2_post_processed", clinic_setname.c), overwrite.l = TRUE)
Proteomics
Set names and input files
proteo_files.vc <- vapply(LATMX2::proteo_sets.vc(), function(set.c) { file.c <- list.files(file.path(LATMX2::processed_dir.c(), set.c), pattern = ".xlsx", full.names = TRUE) stopifnot(length(file.c) == 1) file.c }, FUN.VALUE = character(1)) proteo.mset <- MultiDataSet::createMultiDataSet()
MultiDataSet containing the data matrices
for (set.c in LATMX2::proteo_sets.vc()) { # dataMatrix data.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c], sheet = 1), stringsAsFactors = FALSE) rownames(data.df) <- data.df[, 1] data.df[, 1] <- NULL data.mn <- as.matrix(data.df) rm(data.df) mode(data.mn) <- "numeric" eset <- Biobase::ExpressionSet(assayData = data.mn, experimentData = Biobase::MIAME(title = set.c)) stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
sampleMetadata
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] proteo_pda.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c], sheet = 2), stringsAsFactors = FALSE) rownames(proteo_pda.df) <- gsub(".", "_", proteo_pda.df[, 1], fixed = TRUE) sample_names.vc <- Biobase::sampleNames(eset) if (set.c == "proteomics_liver") { stopifnot(identical(sort(rownames(proteo_pda.df)), sort(sample_names.vc))) proteo_pda.df <- proteo_pda.df[sample_names.vc, ] } stopifnot(identical(rownames(proteo_pda.df), sample_names.vc)) colnames(proteo_pda.df) <- paste0("supp_", colnames(proteo_pda.df)) Biobase::pData(eset) <- proteo_pda.df stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
variableMetadata
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] proteo_fda.df <- as.data.frame(readxl::read_excel(proteo_files.vc[set.c], sheet = 3), stringsAsFactors = FALSE) rownames(proteo_fda.df) <- proteo_fda.df[, 1] stopifnot(identical(rownames(proteo_fda.df), Biobase::featureNames(eset))) proteo_fda.df[, "uniprot_id"] <- sapply(proteo_fda.df[, "accession"], function(access.c) unlist(strsplit(access.c, split = "|", fixed = TRUE))[2], USE.NAMES = FALSE) supp.vi <- which(!(colnames(proteo_fda.df) %in% c("accession", "description", "uniprot_id"))) colnames(proteo_fda.df)[supp.vi] <- paste0("supp_", colnames(proteo_fda.df)[supp.vi]) Biobase::fData(eset) <- proteo_fda.df stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
Renaming features and re-ordering samples
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] # changing variable IDs to names feat_names.vc <- paste0(sapply(Biobase::fData(eset)[, "accession"], function(access.c) unlist(strsplit(access.c, split = "|", fixed = TRUE))[2]), "_", sapply(Biobase::fData(eset)[, "description"], function(desc.c) { if (!is.na(desc.c) && desc.c != "") { desc.c <- unlist(strsplit(desc.c, split = " OS="))[1] if (nchar(desc.c) > 25) desc.c <- paste0(substr(desc.c, 1, 24), ".") return(desc.c) } })) stopifnot(!any(duplicated(feat_names.vc))) Biobase::featureNames(eset) <- feat_names.vc # discarding pools if (set.c == "proteomics_liver") eset <- eset[, !grepl("(p|P)ool", Biobase::pData(eset)[, "supp_sample name"])] # re-ordering samples if (set.c == "proteomics_liver") { eset <- eset[, order(as.integer(Biobase::pData(eset)[, "supp_sample name"]))] } else { Biobase::sampleNames(eset) <- sapply(Biobase::sampleNames(eset), function(samp.c) unlist(strsplit(samp.c, split = "_"))[2], USE.NAMES = FALSE) eset <- eset[, order(as.integer(Biobase::sampleNames(eset)))] } # renaming samples if (set.c == "proteomics_liver") { Biobase::sampleNames(eset) <- paste0(sapply(Biobase::pData(eset)[, "supp_Condition"], function(cond.c) switch(cond.c, mx2 = "X", ctrl = "W", lat = "L")), Biobase::pData(eset)[, "supp_sample name"]) } else { Biobase::sampleNames(eset) <- paste0(sapply(Biobase::pData(eset)[, "supp_Condition"], function(cond.c) switch(cond.c, Mx2 = "X", CONT = "W", LAT = "L")), Biobase::sampleNames(eset)) } mice_konum.vc <- substr(mice_id.vc, 1, 4) mice_koid.vc <- mice_id.vc if (set.c == "proteomics_plasma") { stopifnot(all(Biobase::sampleNames(eset) %in% mice_konum.vc)) mice_koid.vc <- mice_koid.vc[mice_konum.vc %in% Biobase::sampleNames(eset)] mice_konum.vc <- mice_konum.vc[mice_konum.vc %in% Biobase::sampleNames(eset)] } stopifnot(identical(Biobase::sampleNames(eset), mice_konum.vc)) Biobase::sampleNames(eset) <- mice_koid.vc stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
Adding meta data (gene, mouse_nb, sex)
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] Biobase::pData(eset) <- cbind.data.frame(mice_id.df[Biobase::sampleNames(eset), ], Biobase::pData(eset), stringsAsFactors = FALSE) stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
formatting the 'imputation' metrics
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] eset <- ProMetIA::format_imputation(eset) stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
Variable metadata: including new mice ids in the column names (count, abundance, etc.)
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] samp.vc <- Biobase::sampleNames(eset) pdata.df <- Biobase::pData(eset) fvar.vc <- Biobase::fvarLabels(eset) for (i in seq_along(samp.vc)) { samp.c <- samp.vc[i] samp_id.c <- gsub("abundance_", "", pdata.df[i, "id"]) fvar_id.vi <- grep(paste0("^supp_.+", samp_id.c), fvar.vc) fvar.vc[fvar_id.vi] <- gsub(switch(set.c, proteomics_liver = "mgf", proteomics_plasma = samp_id.c), samp.c, fvar.vc[fvar_id.vi]) } Biobase::fvarLabels(eset) <- fvar.vc stopifnot(methods::validObject(eset)) proteo.mset <- MultiDataSet::add_eset(proteo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
Re-ordering metadata ('supp_' columns at the end)
proteo.mset <- ProMetIA::ordermeta_mset(proteo.mset)
Saving (not run)
for (set.c in LATMX2::proteo_sets.vc()) { eset <- proteo.mset[[set.c]] # an 'id' extra column has been automatically created in the sampleMetadata by # 'MultiDataSet' and must be removed before saving pdata.df <- Biobase::pData(eset) pdata.df[, "id"] <- NULL Biobase::pData(eset) <- pdata.df phenomis::writing(eset, file.path("../../LATMX2/inst/extdata/2_post_processed", set.c), overwrite.l = TRUE) }
Metabolomics (in progress)
Loading
metabo.mset <- phenomis::reading(file.path(LATMX2::processed_dir.c()), subsets.vc = LATMX2::metabo_sets.vc(), report.c = "none")
Normalization of the dataMatrix
for (set.c in LATMX2::metabo_sets.vc()) { eset <- metabo.mset[[set.c]] if (grepl("(hyper|hilic)", set.c)) { if (grepl("c18hyper", set.c)) { # setting NA values to 0 exprs.mn <- Biobase::exprs(eset) exprs.mn[is.na(exprs.mn)] <- ifelse(set.c == "metabolomics_liver_c18hyper-pos", 0, 1) Biobase::exprs(eset) <- exprs.mn } # computing quality metrics eset <- phenomis::inspecting(eset, title.c = set.c) # discarding features with mean intensity in samples / mean intensity in pools < 3 eset <- eset[Biobase::fData(eset)[, "blankMean_over_sampleMean"] <= 0.33, ] # discarding features with a Pearson correlation with pool dilution < 0.7 eset <- eset[Biobase::fData(eset)[, "poolDil_cor"] >= 0.7, ] } else if (grepl("acqui", set.c)) { # discarding features eluted before 0.4 min or after 22 min eset <- eset[Biobase::fData(eset)[, "rt"] >= 0.4 & Biobase::fData(eset)[, "rt"] <= 22, ] # discarding features with mean intensity in samples / mean intensity in pools < 2 (or 4) fold.i <- ifelse(set.c == "metabolomics_plasma_c18acqui-pos", 2, 4) eset <- eset[Biobase::fData(eset)[, "fold"] >= fold.i, ] eset <- eset[Biobase::fData(eset)[, "tstat"] <= 0, ] } else stop("Unknown metabolomics dataset name.") # discarding blanks and diluted pools (QCs) eset <- eset[, Biobase::pData(eset)[, "sampleType"] %in% c("sample", "pool")] # signal drift correction if (set.c == "metabolomics_liver_c18hyper-pos") { reference.c <- "none" } else if (set.c == "metabolomics_plasma_c18hyper-pos") { reference.c <- "pool" } else if (set.c == "metabolomics_liver_hilic-neg") { reference.c <- "none" } else if (set.c == "metabolomics_plasma_hilic-neg") { reference.c <- "none" } else if (set.c == "metabolomics_plasma_c18acqui-pos") { reference.c <- "sample" } else if (set.c == "metabolomics_plasma_c18acqui-neg") { reference.c <- "pool" } else stop() if (reference.c != "none") eset <- phenomis::correcting(eset, reference.c = reference.c, title.c = set.c) # updating quality metrics eset <- phenomis::inspecting(eset, title.c = set.c) # discarding features with a coefficient of variation in pools > 30% eset <- eset[Biobase::fData(eset)[, "pool_CV"] <= 0.3, ] if (grepl("acqui", set.c)) { # discarding features with a ratio 'coefficient of variation in samples / coefficient of variation in pools' < 0.8 eset <- eset[Biobase::fData(eset)[, "poolCV_over_sampleCV"] <= 1.25, ] } # discarding 'pool' samples (i.e. QCs) eset <- eset[, Biobase::pData(eset)[, "sampleType"] != "pool"] if (grepl("acqui", set.c)) { # discarding the few features containing NA (generated after the drift correction) eset <- eset[apply(Biobase::exprs(eset), 1, function(feat.vn) !any(is.na(feat.vn))), ] # discarding features highly correlated to others from the same pcgroup with higher mean intensity eset <- ProMetIA::filter_pcgroup(eset) if (set.c == "metabolomics_plasma_c18acqui-pos") { # discarding features with 90% quantile < 5000 eset <- eset[apply(Biobase::exprs(eset), 1, function(feat.vn) quantile(feat.vn, 0.9)) >= 5000, ] } } stopifnot(methods::validObject(eset)) metabo.mset <- MultiDataSet::add_eset(metabo.mset, eset, dataset.type = set.c, GRanges = NA, overwrite = TRUE, warnings = FALSE) }
Formatting and ordering sample names
metabo.mset <- ProMetIA::format_metabonames(metabo.mset, mice_id.df = mice_id.df, mice_id.vc = mice_id.vc, mice_num.vc = mice_num.vc)
Log2 transformation
metabo.mset <- phenomis::transforming(metabo.mset)
Re-ordering metadata ('supp_' columns at the end)
metabo.mset <- ProMetIA::ordermeta_mset(metabo.mset)
Saving (not run)
for (set.c in LATMX2::metabo_sets.vc()) { eset <- metabo.mset[[set.c]] # an 'id' extra column has been automatically created in the sampleMetadata by # 'MultiDataSet' and must be removed before saving pdata.df <- Biobase::pData(eset) pdata.df[, "id"] <- NULL Biobase::pData(eset) <- pdata.df phenomis::writing(eset, file.path("../../LATMX2/inst/extdata/2_post_processed", set.c), overwrite.l = 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.