tests/testthat/test-05_prepare_data.R
In innatelab/msglm: GLM models for Mass Spectrometry Data
context("data preration pipleine")
library(checkmate)
source(test_path("gen-msdata.R"))
# generate simplest model
conditions_df = tibble(infection = factor(c("control", "virus", "control", "virus"), levels=c("control", "virus")),
treatment = factor(c("mock", "mock", "drug", "drug"), levels=c("mock", "drug")),
condition = paste0(infection, "_", treatment))
conditionXeffect <- model.matrix(~ 1 + infection * treatment, data = conditions_df)
conditionXeffect <- conditionXeffect[, -1, drop=FALSE]
rownames(conditionXeffect) <- conditions_df$condition
effects_df = tibble(effect = colnames(conditionXeffect)) %>%
dplyr::filter(effect != "(Intercept)") %>%
dplyr::mutate(is_interaction = str_detect(effect, ':'),
prior_mean = 0,
prior_tau = if_else(is_interaction, 0.5, 1.0),
prior_df1 = if_else(is_interaction, 4, 2),
prior_df2 = if_else(is_interaction, 4, 2))
model_def <- msglm_model(conditionXeffect, conditions_df, effects_df)
mscalib <- read_mscalib_json(test_path("test_mscalib.json")) %>% msglm::convert_logintensityBase(2)
for (obj in c("protgroup", "ptmngroup")) {
for (mstag in c(NA_character_, "mstag")) {
msprobes_df <- tidyr::expand_grid(condition = conditions_df$condition,
replicate = 1:3) %>%
dplyr::mutate(msprobe = paste0(condition, "_", replicate))
if (!is.na(mstag)) {
msprobes_df <- dplyr::mutate(msprobes_df,
!!sym(mstag) := if_else(str_detect(condition, "drug"), "H", "L"),
msexperiment = msprobe,
msprobe = paste0(msexperiment, "_", mstag))
}
for (msprobe in if (is.na(mstag)) c("msprobe", "msexperiment", "msrun", "mschannel") else c("msprobe", "mschannel")) {
msprobe_shifts_df <- tibble(!!sym(msprobe) := msprobes_df$msprobe,
!!sym(paste0("total_", msprobe, "_shift")) := rnorm(nrow(msprobes_df), 0.0, 0.1))
test_that(paste0(obj, "/", obj, " model, no msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
orig_msdata <- gen_msdata(model_def, dplyr::mutate(msprobes_df, mschannel=msprobe),
msprobe = msprobe,
object = obj, nobjects = 3)
msprobes_dfname <- paste0(msprobe, "s")
if (!(msprobe %in% c("msrun", "mschannel"))) {
# add rawfile column to facilitate autodetection
orig_msdata[[msprobes_dfname]] <- dplyr::mutate(orig_msdata[[msprobes_dfname]],
rawfile = paste0(!!sym(msprobe), ".raw"))
}
test_that(paste0("import fails if no ", msprobe, " data provided"), {
bad_msdata <- orig_msdata
bad_msdata[[msprobes_dfname]] <- NULL
expect_error(import_msglm_data(bad_msdata, model_def, mscalib, object=obj),
"Cannot autodetect MS probes")
bad_msdata <- orig_msdata
if (!(msprobe %in% c("msrun", "mschannel"))) {
bad_msdata[[msprobes_dfname]] <- dplyr::mutate(orig_msdata[[msprobes_dfname]],
rawfile = NULL)
expect_error(import_msglm_data(bad_msdata, model_def, mscalib, object=obj),
"Cannot autodetect MS channels")
}
})
if (obj != "protgroup") {
test_that("expecting protgroups by default", {
expect_error(import_msglm_data(orig_msdata, model_def, mscalib=mscalib),
"msdata\\$protgroups not found")
})
}
test_that(paste0("bad ", obj, "_idents frame (", obj, " X ", msprobe, ") is skipped without error"), {
bad_idents_msdata <- orig_msdata
bad_idents_msdata[[paste0(obj, "_idents")]] <- tibble()
msdata <- import_msglm_data(bad_idents_msdata, model_def, object=obj, mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), disjunct.from = paste0(obj, "_idents"))
})
msdata <- import_msglm_data(orig_msdata, model_def, object=obj, mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname,
paste0(obj, "_intensities"), paste0(obj, "s")))
expect_tibble(msdata[[msprobes_dfname]])
expect_names(names(msdata[[msprobes_dfname]]), must.include = c("condition", msprobe))
if (msprobe %in% c("msprobe", "msexperiment")) {
# don't include mschannel/msrun frames
expect_names(names(msdata), disjunct.from = c("msruns", "mschannels"))
}
obj_intens_df <- msdata[[paste0(obj, "_intensities")]]
expect_tibble(obj_intens_df)
expect_names(colnames(obj_intens_df),
must.include = c(paste0(obj, "_id"), msprobe, "intensity"))
expect_equal(msdata$msentities, c(object=obj, quantobject=obj,
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), msprobe, NA_character_), mschannel=msprobe,
mstag=mstag, msfraction=NA_character_, msprotocol=NA_character_))
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
stan_data1 <- to_standata(model_data1)
expect_list(stan_data1)
expect_names(names(stan_data1), must.include=c("Nconditions", "Nobjects",
"Nprobes", "NobjProbes",
"Nmschannels", "mschannel_shift",
"Neffects", "NobjEffects", "NbatchEffects", "NobjBatchEffects",
"NobjConditions", "Nquanted", "Nmissed",
"quant2mschannel", "miss2mschannel",
"quant2obj_probe", "miss2obj_probe", "qData",
"obj_condXeff_Nw", "obj_probeXbatcheff_Nw"))
model_data2 <- msglm_data(model_def, msdata, 2L)
expect_s3_class(model_data2, "msglm_model_data")
expect_equal(model_data2$object_id, 2L)
expect_equal(unique(model_data2$objects$object_id), 2L)
expect_error(msglm_data(model_def, msdata, 5L), "Objects not found")
test_that(paste0(msprobe, "Xeffect matrix could be used for the model"), {
msprobeXeffect <- model.matrix(~ 1 + infection * treatment,
data = dplyr::inner_join(msprobes_df, conditions_df, by="condition"))
msprobeXeffect <- msprobeXeffect[, -1, drop=FALSE]
rownames(msprobeXeffect) <- msprobes_df$msprobe
names(dimnames(msprobeXeffect)) <- c(msprobe, "effect")
msprobe_model_def <- msglm_model(conditionXeffect, conditions_df, effects_df, msprobeXeffect = msprobeXeffect)
model_data3 <- msglm_data(msprobe_model_def, msdata, 2L)
expect_s3_class(model_data3, "msglm_model_data")
})
})
test_that(paste0(obj, "/pepmodstate model, no msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
mschannels_df <- dplyr::mutate(msprobes_df, mschannel = msprobe)
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = msprobe, mschannel = msprobe,
object = obj, quantobject = "pepmodstate", nobjects=3)
msprobes_dfname <- paste0(msprobe, "s")
orig_msdata[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe) %>%
dplyr::mutate(rawfile = paste0(!!sym(msprobe), ".raw"))
if (msprobe != "msprobe") {
orig_msdata[[msprobes_dfname]]$msprobe <- NULL
}
msdata <- import_msglm_data(orig_msdata, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), msprobe, NA_character_), mschannel=msprobe,
mstag=mstag, msfraction=NA_character_, msprotocol=NA_character_))
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "quantobjects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
expect_equal(unique(model_data1$quantobjects$index_object), 1L)
})
}
for (msprobe in if(is.na(mstag)) c("msprobe", "msexperiment") else "msprobe") {
test_that(paste0(obj, "/pepmodstate model, msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
mschannel <- c(msprobe = "mschannel", msexperiment = "msrun")[[msprobe]]
mschannels_df <- dplyr::full_join(msprobes_df, tibble(msfraction = c(1L, 2L)), by=character()) %>%
dplyr::mutate(mschannel = paste0(condition, "_", replicate, "_F", msfraction))
if (!is.na(mstag)) {
mschannels_df <- dplyr::mutate(mschannels_df,
msrun = mschannel,
mschannel = paste0(mschannel, "_", mstag))
}
mschannels_df <- dplyr::mutate(mschannels_df,
rawfile = paste0(condition, '_', replicate,
'_F', msfraction, ".raw"))
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = msprobe, mschannel = mschannel,
object = obj, quantobject = "pepmodstate", nobjects=3)
msprobes_dfname <- paste0(msprobe, "s")
mschannels_dfname <- paste0(mschannel, "s")
# without specifying mschannels, by with hint that msfractions exist
bad_msdata <- orig_msdata
bad_msdata[[paste0(mschannel, "s")]] <- NULL
expect_error(msglm::import_msglm_data(bad_msdata, model_def, mscalib,
object=obj, msfraction="msfraction"),
"Cannot autodetect MS probes")
# without specifying msprobes
orig_msdata1 <- orig_msdata
msdata <- import_msglm_data(orig_msdata1, model_def,
object=obj, quantobject="pepmodstate",
mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname, mschannels_dfname,
"pepmodstate_intensities", paste0(obj, "s"), "pepmodstates",
paste0(obj, "2pepmodstate")))
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol = NA_character_))
expect_tibble(msdata[[msprobes_dfname]], nrows=nrow(msprobes_df))
expect_names(names(msdata[[msprobes_dfname]]), must.include = c("condition", msprobe))
expect_tibble(msdata[[mschannels_dfname]], nrows=nrow(mschannels_df))
expect_names(names(msdata[[mschannels_dfname]]), must.include = c(msprobe, mschannel))
expect_tibble(msdata$pepmodstate_intensities)
expect_names(names(msdata$pepmodstate_intensities),
must.include = c("pepmodstate_id", mschannel, "intensity"))
# without specifying msprobes and without msprobe column in mschannels
orig_msdata1b <- orig_msdata
orig_msdata1b[[mschannels_dfname]] <- dplyr::mutate(orig_msdata1b[[mschannels_dfname]], !!sym(msprobe) := NULL)
msdata <- msglm::import_msglm_data(orig_msdata1b, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname, mschannels_dfname,
"pepmodstate_intensities",
paste0(obj, "s"), "pepmodstates", paste0(obj, "2pepmodstate")))
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol=NA_character_))
expect_tibble(msdata[[msprobes_dfname]], nrows=nrow(msprobes_df))
expect_tibble(msdata[[mschannels_dfname]], nrows=nrow(mschannels_df))
# with obj2pepmod but without obj2pepmodstate
orig_msdata1c <- orig_msdata
obj2pms_dfname <- paste0(obj, "2pepmodstate")
orig_msdata1c[[obj2pms_dfname]] <- NULL
msdata <- msglm::import_msglm_data(orig_msdata1c, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
# check that obj2pepmodstate is automatically generated
expect_names(names(msdata), must.include = obj2pms_dfname)
expect_tibble(msdata[[obj2pms_dfname]], nrows=nrow(orig_msdata[[obj2pms_dfname]]))
expect_names(names(msdata[[obj2pms_dfname]]),
must.include = c(paste0(c(obj, "pepmodstate"), "_id"), "is_specific"))
# without specifying mschannels
orig_msdata2 <- orig_msdata
orig_msdata2[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe)
orig_msdata2[[paste0(mschannel, "s")]] <- NULL
expect_error(msglm::import_msglm_data(orig_msdata2, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib),
"Cannot autodetect MS channels data")
# with both
orig_msdata3 <- orig_msdata
orig_msdata3[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe)
msdata <- import_msglm_data(orig_msdata3, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol=NA_character_))
if (mschannel != msprobe) {
msdata[[paste0(mschannel, "_shifts")]] <- tibble(
!!sym(mschannel) := mschannels_df$mschannel,
!!sym(paste0("total_", mschannel, "_shift")) := rnorm(nrow(mschannels_df), 0.0, 0.1))
} else {
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
}
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "quantobjects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
expect_equal(unique(model_data1$quantobjects$object_id), 1L)
stan_data1 <- to_standata(model_data1)
expect_list(stan_data1)
expect_names(names(stan_data1), must.include=c(
"Nconditions", "Nobjects", "Nquantobjects",
"Nprobes", "NobjProbes", "NqobjChannels",
"Nmschannels", "mschannel_shift",
"Neffects", "NobjEffects", "NbatchEffects", "NobjBatchEffects",
"NquantBatchEffects", "NqobjBatchEffects",
"NobjConditions", "Nquanted", "Nmissed",
"quant2qobj_channel", "miss2qobj_channel", "qData",
"obj_condXeff_Nw", "obj_probeXbatcheff_Nw", "qobj_channelXqbatcheff_Nw"))
})
}}}
test_that("Technical MS replicates are supported", {
msprobes_df <- tidyr::expand_grid(condition = conditions_df$condition,
replicate = 1:4) %>%
dplyr::mutate(msprobe = paste0(condition, "_", replicate))
mschannels_df <- tidyr::expand_grid(msprobe = msprobes_df$msprobe,
tech_replicate = 1:2) %>%
dplyr::left_join(dplyr::select(msprobes_df, msprobe, condition), by="msprobe") %>%
dplyr::mutate(mschannel = paste0(msprobe, "_", tech_replicate))
mschannel_shifts_df <- tibble(mschannel = mschannels_df$mschannel,
total_mschannel_shift = rnorm(nrow(mschannels_df), 0.0, 0.1))
test_that("Technical MS replicates are supported on pepmodstate level", {
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = "msprobe", mschannel = "mschannel",
object = "protgroup", quantobject = "pepmodstate", nobjects=3)
msdata <- import_msglm_data(orig_msdata, model_def,
object="protgroup", quantobject="pepmodstate",
mscalib=mscalib)
msdata$mschannel_shifts <- mschannel_shifts_df
model_data <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data, "msglm_model_data")
expect_equal(nrow(model_data$msprobes), nrow(msprobes_df))
expect_equal(nrow(model_data$mschannels), nrow(mschannels_df))
stan_data <- to_standata(model_data)
expect_list(stan_data)
test_that("Technical MS replicates + quant batch effects work", {
mschannels_be_df <- dplyr::mutate(dplyr::left_join(mschannels_df, dplyr::select(msprobes_df, msprobe, replicate),
by="msprobe"),
batch = LETTERS[pmin(3L, 1L + (replicate + tech_replicate) %% 4)])
batch_effect_mtx <- model.matrix(~ 1 + batch, data = mschannels_be_df)
batch_effect_mtx <- batch_effect_mtx[, -1, drop=FALSE]
dimnames(batch_effect_mtx) <- list(mschannel = mschannels_be_df$mschannel,
quant_batch_effect = colnames(batch_effect_mtx))
model_def_be <- set_batch_effects(model_def, batch_effect_mtx, applies_to = "quantobject")
expect_matrix(model_def_be$mschannelXquantBatchEffect, any.missing = FALSE,
nrows=nrow(mschannels_be_df), ncols=n_distinct(mschannels_be_df$batch) - 1L)
expect_data_frame(model_def_be$quant_batch_effects, nrows=n_distinct(mschannels_be_df$batch) - 1L)
model_data_be <- msglm_data(model_def_be, msdata, 1L)
expect_data_frame(model_data_be$quantobject_batch_effects)
expect_names(names(model_data_be$quantobject_batch_effects),
must.include = c("quantobject_batch_effect", "quant_batch_effect", "index_quantobject"))
expect_matrix(model_data_be$quantobject_mschannelXquant_batch_effect, any.missing = FALSE,
nrows = nrow(model_data_be$msdata),
ncols = nrow(model_data_be$quantobject_batch_effects))
stan_data_be <- to_standata(model_data_be)
expect_equal(stan_data_be$NquantBatchEffects, n_distinct(mschannels_be_df$batch) - 1L)
expect_equal(stan_data_be$NqobjBatchEffects, nrow(model_data_be$quantobject_batch_effects))
})
})
test_that("Technical MS replicates are supported on protgroup level", {
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = "msprobe", mschannel = "mschannel",
object = "protgroup", quantobject = NA_character_, nobjects=3)
msdata <- import_msglm_data(orig_msdata, model_def,
object="protgroup", quantobject="protgroup",
mscalib=mscalib, verbose=TRUE)
msdata$mschannel_shifts <- mschannel_shifts_df
model_data <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data, "msglm_model_data")
expect_data_frame(model_data$msprobes, nrows=nrow(msprobes_df))
expect_data_frame(model_data$mschannels, nrows=nrow(mschannels_df))
stan_data <- to_standata(model_data)
expect_list(stan_data)
expect_equal(stan_data$Nprobes, nrow(msprobes_df))
expect_equal(stan_data$Nmschannels, nrow(mschannels_df))
expect_integer(stan_data$quant2mschannel,
lower=1, upper=stan_data$Nmschannels,
len=sum(model_data$msdata$is_observed))
expect_integer(stan_data$miss2mschannel,
lower=1, upper=stan_data$Nmschannels,
len=sum(!model_data$msdata$is_observed))
})
})
innatelab/msglm documentation built on March 23, 2023, 7:17 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.
context("data preration pipleine")
library(checkmate)
source(test_path("gen-msdata.R"))
# generate simplest model
conditions_df = tibble(infection = factor(c("control", "virus", "control", "virus"), levels=c("control", "virus")),
treatment = factor(c("mock", "mock", "drug", "drug"), levels=c("mock", "drug")),
condition = paste0(infection, "_", treatment))
conditionXeffect <- model.matrix(~ 1 + infection * treatment, data = conditions_df)
conditionXeffect <- conditionXeffect[, -1, drop=FALSE]
rownames(conditionXeffect) <- conditions_df$condition
effects_df = tibble(effect = colnames(conditionXeffect)) %>%
dplyr::filter(effect != "(Intercept)") %>%
dplyr::mutate(is_interaction = str_detect(effect, ':'),
prior_mean = 0,
prior_tau = if_else(is_interaction, 0.5, 1.0),
prior_df1 = if_else(is_interaction, 4, 2),
prior_df2 = if_else(is_interaction, 4, 2))
model_def <- msglm_model(conditionXeffect, conditions_df, effects_df)
mscalib <- read_mscalib_json(test_path("test_mscalib.json")) %>% msglm::convert_logintensityBase(2)
for (obj in c("protgroup", "ptmngroup")) {
for (mstag in c(NA_character_, "mstag")) {
msprobes_df <- tidyr::expand_grid(condition = conditions_df$condition,
replicate = 1:3) %>%
dplyr::mutate(msprobe = paste0(condition, "_", replicate))
if (!is.na(mstag)) {
msprobes_df <- dplyr::mutate(msprobes_df,
!!sym(mstag) := if_else(str_detect(condition, "drug"), "H", "L"),
msexperiment = msprobe,
msprobe = paste0(msexperiment, "_", mstag))
}
for (msprobe in if (is.na(mstag)) c("msprobe", "msexperiment", "msrun", "mschannel") else c("msprobe", "mschannel")) {
msprobe_shifts_df <- tibble(!!sym(msprobe) := msprobes_df$msprobe,
!!sym(paste0("total_", msprobe, "_shift")) := rnorm(nrow(msprobes_df), 0.0, 0.1))
test_that(paste0(obj, "/", obj, " model, no msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
orig_msdata <- gen_msdata(model_def, dplyr::mutate(msprobes_df, mschannel=msprobe),
msprobe = msprobe,
object = obj, nobjects = 3)
msprobes_dfname <- paste0(msprobe, "s")
if (!(msprobe %in% c("msrun", "mschannel"))) {
# add rawfile column to facilitate autodetection
orig_msdata[[msprobes_dfname]] <- dplyr::mutate(orig_msdata[[msprobes_dfname]],
rawfile = paste0(!!sym(msprobe), ".raw"))
}
test_that(paste0("import fails if no ", msprobe, " data provided"), {
bad_msdata <- orig_msdata
bad_msdata[[msprobes_dfname]] <- NULL
expect_error(import_msglm_data(bad_msdata, model_def, mscalib, object=obj),
"Cannot autodetect MS probes")
bad_msdata <- orig_msdata
if (!(msprobe %in% c("msrun", "mschannel"))) {
bad_msdata[[msprobes_dfname]] <- dplyr::mutate(orig_msdata[[msprobes_dfname]],
rawfile = NULL)
expect_error(import_msglm_data(bad_msdata, model_def, mscalib, object=obj),
"Cannot autodetect MS channels")
}
})
if (obj != "protgroup") {
test_that("expecting protgroups by default", {
expect_error(import_msglm_data(orig_msdata, model_def, mscalib=mscalib),
"msdata\\$protgroups not found")
})
}
test_that(paste0("bad ", obj, "_idents frame (", obj, " X ", msprobe, ") is skipped without error"), {
bad_idents_msdata <- orig_msdata
bad_idents_msdata[[paste0(obj, "_idents")]] <- tibble()
msdata <- import_msglm_data(bad_idents_msdata, model_def, object=obj, mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), disjunct.from = paste0(obj, "_idents"))
})
msdata <- import_msglm_data(orig_msdata, model_def, object=obj, mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname,
paste0(obj, "_intensities"), paste0(obj, "s")))
expect_tibble(msdata[[msprobes_dfname]])
expect_names(names(msdata[[msprobes_dfname]]), must.include = c("condition", msprobe))
if (msprobe %in% c("msprobe", "msexperiment")) {
# don't include mschannel/msrun frames
expect_names(names(msdata), disjunct.from = c("msruns", "mschannels"))
}
obj_intens_df <- msdata[[paste0(obj, "_intensities")]]
expect_tibble(obj_intens_df)
expect_names(colnames(obj_intens_df),
must.include = c(paste0(obj, "_id"), msprobe, "intensity"))
expect_equal(msdata$msentities, c(object=obj, quantobject=obj,
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), msprobe, NA_character_), mschannel=msprobe,
mstag=mstag, msfraction=NA_character_, msprotocol=NA_character_))
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
stan_data1 <- to_standata(model_data1)
expect_list(stan_data1)
expect_names(names(stan_data1), must.include=c("Nconditions", "Nobjects",
"Nprobes", "NobjProbes",
"Nmschannels", "mschannel_shift",
"Neffects", "NobjEffects", "NbatchEffects", "NobjBatchEffects",
"NobjConditions", "Nquanted", "Nmissed",
"quant2mschannel", "miss2mschannel",
"quant2obj_probe", "miss2obj_probe", "qData",
"obj_condXeff_Nw", "obj_probeXbatcheff_Nw"))
model_data2 <- msglm_data(model_def, msdata, 2L)
expect_s3_class(model_data2, "msglm_model_data")
expect_equal(model_data2$object_id, 2L)
expect_equal(unique(model_data2$objects$object_id), 2L)
expect_error(msglm_data(model_def, msdata, 5L), "Objects not found")
test_that(paste0(msprobe, "Xeffect matrix could be used for the model"), {
msprobeXeffect <- model.matrix(~ 1 + infection * treatment,
data = dplyr::inner_join(msprobes_df, conditions_df, by="condition"))
msprobeXeffect <- msprobeXeffect[, -1, drop=FALSE]
rownames(msprobeXeffect) <- msprobes_df$msprobe
names(dimnames(msprobeXeffect)) <- c(msprobe, "effect")
msprobe_model_def <- msglm_model(conditionXeffect, conditions_df, effects_df, msprobeXeffect = msprobeXeffect)
model_data3 <- msglm_data(msprobe_model_def, msdata, 2L)
expect_s3_class(model_data3, "msglm_model_data")
})
})
test_that(paste0(obj, "/pepmodstate model, no msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
mschannels_df <- dplyr::mutate(msprobes_df, mschannel = msprobe)
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = msprobe, mschannel = msprobe,
object = obj, quantobject = "pepmodstate", nobjects=3)
msprobes_dfname <- paste0(msprobe, "s")
orig_msdata[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe) %>%
dplyr::mutate(rawfile = paste0(!!sym(msprobe), ".raw"))
if (msprobe != "msprobe") {
orig_msdata[[msprobes_dfname]]$msprobe <- NULL
}
msdata <- import_msglm_data(orig_msdata, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), msprobe, NA_character_), mschannel=msprobe,
mstag=mstag, msfraction=NA_character_, msprotocol=NA_character_))
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "quantobjects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
expect_equal(unique(model_data1$quantobjects$index_object), 1L)
})
}
for (msprobe in if(is.na(mstag)) c("msprobe", "msexperiment") else "msprobe") {
test_that(paste0(obj, "/pepmodstate model, msfractions, ",
if_else(!is.na(mstag), "", "no "), "mstags, specifying ", msprobe, "s"), {
mschannel <- c(msprobe = "mschannel", msexperiment = "msrun")[[msprobe]]
mschannels_df <- dplyr::full_join(msprobes_df, tibble(msfraction = c(1L, 2L)), by=character()) %>%
dplyr::mutate(mschannel = paste0(condition, "_", replicate, "_F", msfraction))
if (!is.na(mstag)) {
mschannels_df <- dplyr::mutate(mschannels_df,
msrun = mschannel,
mschannel = paste0(mschannel, "_", mstag))
}
mschannels_df <- dplyr::mutate(mschannels_df,
rawfile = paste0(condition, '_', replicate,
'_F', msfraction, ".raw"))
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = msprobe, mschannel = mschannel,
object = obj, quantobject = "pepmodstate", nobjects=3)
msprobes_dfname <- paste0(msprobe, "s")
mschannels_dfname <- paste0(mschannel, "s")
# without specifying mschannels, by with hint that msfractions exist
bad_msdata <- orig_msdata
bad_msdata[[paste0(mschannel, "s")]] <- NULL
expect_error(msglm::import_msglm_data(bad_msdata, model_def, mscalib,
object=obj, msfraction="msfraction"),
"Cannot autodetect MS probes")
# without specifying msprobes
orig_msdata1 <- orig_msdata
msdata <- import_msglm_data(orig_msdata1, model_def,
object=obj, quantobject="pepmodstate",
mscalib=mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname, mschannels_dfname,
"pepmodstate_intensities", paste0(obj, "s"), "pepmodstates",
paste0(obj, "2pepmodstate")))
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol = NA_character_))
expect_tibble(msdata[[msprobes_dfname]], nrows=nrow(msprobes_df))
expect_names(names(msdata[[msprobes_dfname]]), must.include = c("condition", msprobe))
expect_tibble(msdata[[mschannels_dfname]], nrows=nrow(mschannels_df))
expect_names(names(msdata[[mschannels_dfname]]), must.include = c(msprobe, mschannel))
expect_tibble(msdata$pepmodstate_intensities)
expect_names(names(msdata$pepmodstate_intensities),
must.include = c("pepmodstate_id", mschannel, "intensity"))
# without specifying msprobes and without msprobe column in mschannels
orig_msdata1b <- orig_msdata
orig_msdata1b[[mschannels_dfname]] <- dplyr::mutate(orig_msdata1b[[mschannels_dfname]], !!sym(msprobe) := NULL)
msdata <- msglm::import_msglm_data(orig_msdata1b, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_names(names(msdata), must.include = c("msentities", msprobes_dfname, mschannels_dfname,
"pepmodstate_intensities",
paste0(obj, "s"), "pepmodstates", paste0(obj, "2pepmodstate")))
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol=NA_character_))
expect_tibble(msdata[[msprobes_dfname]], nrows=nrow(msprobes_df))
expect_tibble(msdata[[mschannels_dfname]], nrows=nrow(mschannels_df))
# with obj2pepmod but without obj2pepmodstate
orig_msdata1c <- orig_msdata
obj2pms_dfname <- paste0(obj, "2pepmodstate")
orig_msdata1c[[obj2pms_dfname]] <- NULL
msdata <- msglm::import_msglm_data(orig_msdata1c, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
# check that obj2pepmodstate is automatically generated
expect_names(names(msdata), must.include = obj2pms_dfname)
expect_tibble(msdata[[obj2pms_dfname]], nrows=nrow(orig_msdata[[obj2pms_dfname]]))
expect_names(names(msdata[[obj2pms_dfname]]),
must.include = c(paste0(c(obj, "pepmodstate"), "_id"), "is_specific"))
# without specifying mschannels
orig_msdata2 <- orig_msdata
orig_msdata2[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe)
orig_msdata2[[paste0(mschannel, "s")]] <- NULL
expect_error(msglm::import_msglm_data(orig_msdata2, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib),
"Cannot autodetect MS channels data")
# with both
orig_msdata3 <- orig_msdata
orig_msdata3[[msprobes_dfname]] <- dplyr::rename(msprobes_df, !!sym(msprobe) := msprobe)
msdata <- import_msglm_data(orig_msdata3, model_def,
object = obj, quantobject = "pepmodstate",
mscalib = mscalib)
expect_s3_class(msdata, "msglm_data_collection")
expect_equal(msdata$msentities, c(object=obj, quantobject="pepmodstate",
condition="condition",
msexperiment=if_else(is.na(mstag), msprobe, NA_character_), msprobe=msprobe,
msrun=if_else(is.na(mstag), mschannel, NA_character_), mschannel=mschannel,
mstag=mstag, msfraction="msfraction", msprotocol=NA_character_))
if (mschannel != msprobe) {
msdata[[paste0(mschannel, "_shifts")]] <- tibble(
!!sym(mschannel) := mschannels_df$mschannel,
!!sym(paste0("total_", mschannel, "_shift")) := rnorm(nrow(mschannels_df), 0.0, 0.1))
} else {
msdata[[paste0(msprobe, "_shifts")]] <- msprobe_shifts_df
}
model_data1 <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data1, "msglm_model_data")
expect_names(names(model_data1), must.include = c("model_def", "object_id", "msentities",
"objects", "quantobjects", "msprobes", "mschannels"))
expect_tibble(model_data1$objects)
expect_equal(model_data1$object_id, 1L)
expect_equal(unique(model_data1$objects$object_id), 1L)
expect_equal(unique(model_data1$quantobjects$object_id), 1L)
stan_data1 <- to_standata(model_data1)
expect_list(stan_data1)
expect_names(names(stan_data1), must.include=c(
"Nconditions", "Nobjects", "Nquantobjects",
"Nprobes", "NobjProbes", "NqobjChannels",
"Nmschannels", "mschannel_shift",
"Neffects", "NobjEffects", "NbatchEffects", "NobjBatchEffects",
"NquantBatchEffects", "NqobjBatchEffects",
"NobjConditions", "Nquanted", "Nmissed",
"quant2qobj_channel", "miss2qobj_channel", "qData",
"obj_condXeff_Nw", "obj_probeXbatcheff_Nw", "qobj_channelXqbatcheff_Nw"))
})
}}}
test_that("Technical MS replicates are supported", {
msprobes_df <- tidyr::expand_grid(condition = conditions_df$condition,
replicate = 1:4) %>%
dplyr::mutate(msprobe = paste0(condition, "_", replicate))
mschannels_df <- tidyr::expand_grid(msprobe = msprobes_df$msprobe,
tech_replicate = 1:2) %>%
dplyr::left_join(dplyr::select(msprobes_df, msprobe, condition), by="msprobe") %>%
dplyr::mutate(mschannel = paste0(msprobe, "_", tech_replicate))
mschannel_shifts_df <- tibble(mschannel = mschannels_df$mschannel,
total_mschannel_shift = rnorm(nrow(mschannels_df), 0.0, 0.1))
test_that("Technical MS replicates are supported on pepmodstate level", {
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = "msprobe", mschannel = "mschannel",
object = "protgroup", quantobject = "pepmodstate", nobjects=3)
msdata <- import_msglm_data(orig_msdata, model_def,
object="protgroup", quantobject="pepmodstate",
mscalib=mscalib)
msdata$mschannel_shifts <- mschannel_shifts_df
model_data <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data, "msglm_model_data")
expect_equal(nrow(model_data$msprobes), nrow(msprobes_df))
expect_equal(nrow(model_data$mschannels), nrow(mschannels_df))
stan_data <- to_standata(model_data)
expect_list(stan_data)
test_that("Technical MS replicates + quant batch effects work", {
mschannels_be_df <- dplyr::mutate(dplyr::left_join(mschannels_df, dplyr::select(msprobes_df, msprobe, replicate),
by="msprobe"),
batch = LETTERS[pmin(3L, 1L + (replicate + tech_replicate) %% 4)])
batch_effect_mtx <- model.matrix(~ 1 + batch, data = mschannels_be_df)
batch_effect_mtx <- batch_effect_mtx[, -1, drop=FALSE]
dimnames(batch_effect_mtx) <- list(mschannel = mschannels_be_df$mschannel,
quant_batch_effect = colnames(batch_effect_mtx))
model_def_be <- set_batch_effects(model_def, batch_effect_mtx, applies_to = "quantobject")
expect_matrix(model_def_be$mschannelXquantBatchEffect, any.missing = FALSE,
nrows=nrow(mschannels_be_df), ncols=n_distinct(mschannels_be_df$batch) - 1L)
expect_data_frame(model_def_be$quant_batch_effects, nrows=n_distinct(mschannels_be_df$batch) - 1L)
model_data_be <- msglm_data(model_def_be, msdata, 1L)
expect_data_frame(model_data_be$quantobject_batch_effects)
expect_names(names(model_data_be$quantobject_batch_effects),
must.include = c("quantobject_batch_effect", "quant_batch_effect", "index_quantobject"))
expect_matrix(model_data_be$quantobject_mschannelXquant_batch_effect, any.missing = FALSE,
nrows = nrow(model_data_be$msdata),
ncols = nrow(model_data_be$quantobject_batch_effects))
stan_data_be <- to_standata(model_data_be)
expect_equal(stan_data_be$NquantBatchEffects, n_distinct(mschannels_be_df$batch) - 1L)
expect_equal(stan_data_be$NqobjBatchEffects, nrow(model_data_be$quantobject_batch_effects))
})
})
test_that("Technical MS replicates are supported on protgroup level", {
orig_msdata <- gen_msdata(model_def, mschannels_df, msprobe = "msprobe", mschannel = "mschannel",
object = "protgroup", quantobject = NA_character_, nobjects=3)
msdata <- import_msglm_data(orig_msdata, model_def,
object="protgroup", quantobject="protgroup",
mscalib=mscalib, verbose=TRUE)
msdata$mschannel_shifts <- mschannel_shifts_df
model_data <- msglm_data(model_def, msdata, 1L)
expect_s3_class(model_data, "msglm_model_data")
expect_data_frame(model_data$msprobes, nrows=nrow(msprobes_df))
expect_data_frame(model_data$mschannels, nrows=nrow(mschannels_df))
stan_data <- to_standata(model_data)
expect_list(stan_data)
expect_equal(stan_data$Nprobes, nrow(msprobes_df))
expect_equal(stan_data$Nmschannels, nrow(mschannels_df))
expect_integer(stan_data$quant2mschannel,
lower=1, upper=stan_data$Nmschannels,
len=sum(model_data$msdata$is_observed))
expect_integer(stan_data$miss2mschannel,
lower=1, upper=stan_data$Nmschannels,
len=sum(!model_data$msdata$is_observed))
})
})
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.