tests/testthat/test_netConstruct.R
In stefpeschel/NetCoMi: Network Construction and Comparison for Microbiome Data
set.seed(123456)
data("amgut1.filt")
data("amgut2.filt.phy")
context("netConstruct with phyloseq object as input")
library(phyloseq)
testnet <- netConstruct(amgut2.filt.phy,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext("phyloseq", side = 3, cex = 1.5)
#===============================================================================
# Filtering
context("netConstruct with different taxa filtering methods (single network)")
set.seed(123456)
ftax <- c("none", "totalReads", "relFreq", "numbSamp", "highestVar", "highestFreq")
ftaxpar <- c(list(totalReads = 1000), list(relFreq = 0.05),
list(numbSamp = 10), list(highestVar = 50), list(highestFreq = 50))
dims <- c(127, 125, 3, 127, 50, 50)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
#-------------------------------------------------------------------------------
context("netConstruct with different sample filtering methods (single network")
set.seed(123456)
fsamp <- c("none", "totalReads", "numbTaxa", "highestFreq")
fsampar <- c(list(totalReads = 1000), list(numbTaxa = 50),list(highestFreq = 50))
dims <- c(289, 289, 282, 50)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims[i])
}
#===============================================================================
# Association / dissimilarity measures
context("netConstruct with different association measures")
set.seed(123456)
measures <- c("pearson", "spearman", "bicor", "sparcc", "cclasso", "ccrepe",
"propr","gcoda", "spieceasi_gl", "spieceasi_mb", "spring" )
for (i in 1:length(measures)) {
context(measures[i])
measure.tmp <- measures[i]
if (measure.tmp == "spieceasi_gl") {
measurePar <- list(method = "glasso",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.2))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spieceasi_mb") {
measurePar <- list(method = "mb",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.2))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spring") {
measurePar <- list(nlambda = 5, rep.num = 5)
} else if (measure.tmp == "gcoda") {
measurePar <- list(nlambda = 5)
} else {
measurePar <- NULL
}
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = measure.tmp,
measurePar = measurePar,
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
}
context("test plot.microNetProps")
plot(testprops)
context("netConstruct with association matrix as input")
set.seed(123456)
testnet1 <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testthat::expect_that(netConstruct(testnet1$assoEst1, dataType = "correlation",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)$assoMat1,
testthat::equals(testnet1$assoMat1))
testthat::expect_that(netConstruct(testnet1$assoEst1, dataType = "correlation",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)$adjaMat1,
testthat::equals(testnet1$adjaMat1))
testnet1 <- netConstruct(amgut1.filt,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 30),
zeroMethod = "none", normMethod = "none",
measure = "ckld",
sparsMethod = "knn", seed = 20190101)
testthat::expect_that(netConstruct(testnet1$dissEst1, dataType = "dissimilarity",
sparsMethod = "knn", seed = 20190101)$dissMat1,
testthat::equals(testnet1$dissMat1))
testthat::expect_that(netConstruct(testnet1$dissEst1, dataType = "dissimilarity",
sparsMethod = "knn", seed = 20190101)$adjaMat1,
testthat::equals(testnet1$adjaMat1))
#-------------------------------------------------------------------------------
context("netConstruct with different dissimilarity measures")
set.seed(123456)
measures <- c("euclidean", "bray", "kld", "jeffrey", "ckld",
"jsd", "aitchison")
for (i in 1:length(measures)) {
context(measures[i])
testnet1 <- netConstruct(amgut1.filt,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 30),
zeroMethod = "none", normMethod = "none",
measure = measures[i],
sparsMethod = "knn", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet1, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
}
context("test plot.microNetProps")
plot(testprops)
#===============================================================================
# Zero handling
context("netConstruct with different methods for zero replacement")
set.seed(123456)
zeroMethod <- c("none", "pseudo", "pseudoZO", "multRepl", "alrEM", "bayesMult")
for (i in 1:length(zeroMethod)) {
context(zeroMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[i], normMethod = "clr",
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(zeroMethod[i], side = 3, cex = 1.5)
}
#===============================================================================
# Normalization + zero handling
context("netConstruct with different normalization methods")
set.seed(123456)
normMethod <- c("none", "fractions", "TSS", "CSS", "COM", "rarefy", "clr", "mclr")
zeroMethod <- c("none", "pseudo", "multRepl")
for (i in 1:length(normMethod)) {
for (z in 1:length(zeroMethod)) {
# context(paste0("normMethod: ", normMethod[i], "; zeroMethod: ", zeroMethod[z]))
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[z],
normMethod = normMethod[i],
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(paste0(normMethod[i], ", ", zeroMethod[z]), side = 3, cex = 1.5)
}
}
#===============================================================================
# Sparsification
context("netConstruct with different sparsification methods")
set.seed(123456)
sparsMethod <- c("none", "t-test", "bootstrap",
"threshold", "softThreshold")
for (i in 1:length(sparsMethod)) {
context(sparsMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 100),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson", adjust = "adaptBH",
sparsMethod = sparsMethod[i],
thresh = 0.3, softThreshPower = 10,
dissFunc = "signed", nboot = 5, cores = 1,
seed = 20190101, logFile = NULL,
verbose = FALSE)
if (sparsMethod[i] != "bootstrap") {
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(sparsMethod[i], side = 3, cex = 1.5)
}
}
#===============================================================================
# Multiple testing adjustment
context("netConstruct with different adjustment methods")
set.seed(123456)
adjustMethod <- c("lfdr", "holm", "BH", "BY", rep("adaptBH", 5))
trueNullMethod <- c(rep("convest", 5), "lfdr", "mean", "hist", "farco")
for (i in 1:length(adjustMethod)) {
context(adjustMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "t-test",
adjust = adjustMethod[i],
lfdrThresh = 0.2,
trueNullMethod = trueNullMethod[i],
dissFunc = "signed", nboot = 1000,
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(adjustMethod[i], side = 3, cex = 1.5)
}
#===============================================================================
# Soft-thresholding
context("netConstruct with different values for softThreshType")
set.seed(123456)
softThreshType <- c("signed", "unsigned", "signed hybrid")
for (i in 1:length(softThreshType)) {
context(softThreshType[i])
if (i == 1) {
softThreshPower <- 5
} else {
softThreshPower <- NULL
}
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "softThreshold",
softThreshType = softThreshType[i],
softThreshPower = softThreshPower,
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(softThreshType[i], side = 3, cex = 1.5)
}
#===============================================================================
# K-nearest neighbor sparsification
context("netConstruct with knn sparsification")
set.seed(123456)
knnMutual <- c(TRUE, FALSE)
for (i in 1:length(knnMutual)) {
context(knnMutual[i])
testnet <- netConstruct(amgut1.filt,
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 50),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "knn",
knnMutual = knnMutual[i],
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(knnMutual[i], side = 3, cex = 1.5)
}
#===============================================================================
# Dissimilarity function
context("netConstruct with differenct values for dissFunc")
set.seed(123456)
dissFunc <- c("signed", "unsigned", "signedPos", "TOMdiss")
for (i in 1:length(dissFunc)) {
context(dissFunc[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = dissFunc[i],
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(dissFunc[i], side = 3, cex = 1.5)
}
#-------------------------------------------------------------------------------
# 'dissFunc' is a function
testfunc <- function(x) {
xvec <- x[lower.tri(x)]
dissvec <- sqrt(0.5 * (1-xvec))
dissMat <- x
dissMat[lower.tri(dissMat)] <- dissvec
dissMat[upper.tri(dissMat)] <- t(dissMat)[upper.tri(t(dissMat))]
dissMat[x == 0] <- Inf
diag(dissMat) <- 0
return(dissMat)
}
context("'dissFunc' is a function")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = testfunc,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext("dissFunc as function", side = 3, cex = 1.5)
#===============================================================================
# Similarity function
testfunc <- function(x, power) {
1/(1 + x^power)
}
context("simFunc")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = "signed",
simFunc = testfunc,
simFuncPar = list(2),
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext("simFunc", side = 3, cex = 1.5)
#===============================================================================
# scaleDiss
context("scaleDiss")
scaleDiss <- c(TRUE, FALSE)
for (i in 1:length(scaleDiss)) {
context(scaleDiss[i])
testnet <- netConstruct(amgut1.filt,
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 50),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "knn",
dissFunc = "signed",
scaleDiss = scaleDiss[i],
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(scaleDiss[i], side = 3, cex = 1.5)
}
#===============================================================================
# Unweighted network
context("unweighted network")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
weighted = FALSE,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
summary(testprops)
plot(testprops)
mtext("weighted", side = 3, cex = 1.5)
################################################################################
### two networks
################################################################################
set.seed(123456)
groups_diss <- sample(0:1, ncol(amgut1.filt), replace = TRUE)
groups_asso <- sample(0:1, nrow(amgut1.filt), replace = TRUE)
amgut_male <- phyloseq::subset_samples(amgut2.filt.phy, SEX == "male")
amgut_female <- phyloseq::subset_samples(amgut2.filt.phy, SEX == "female")
context("netConstruct with group vector (association)")
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with data and data2 (association)")
testnet <- netConstruct(amgut_male, amgut_female,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with group vector (dissimilarity)")
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with data and data2 (dissimilarity)")
amgut_diss1 <- amgut1.filt[, groups_diss == 0]
amgut_diss2 <- amgut1.filt[, groups_diss == 1]
testnet <- netConstruct(amgut_diss1, data2 = amgut_diss2,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
#===============================================================================
# Matched-group designs
context("netConstruct with matched-group design")
set.seed(123456)
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:280, ],
matchDesign = c(1,1),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:281, ],
matchDesign = c(1,1),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
testnet <- netConstruct(amgut1.filt[1:95, ], amgut1.filt[96:285, ],
matchDesign = c(1,2),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:280, ],
matchDesign = c(1,2),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
#===============================================================================
# jointPrepro
context("netConstruct with jointPrepro")
set.seed(123456)
testnet <- netConstruct(amgut1.filt, group = groups_asso,
jointPrepro = FALSE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testnet <- netConstruct(amgut_male, amgut_female,
jointPrepro = TRUE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt, group = groups_diss,
jointPrepro = TRUE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
#===============================================================================
# Filtering
##context("netConstruct with different taxa filtering methods (two networks)")
set.seed(123456)
ftax <- c("none", "totalReads", "relFreq", "numbSamp", "highestVar", "highestFreq")
ftaxpar <- c(list(totalReads = 1000), list(relFreq = 0.05),
list(numbSamp = 10), list(highestVar = 50), list(highestFreq = 50))
dims <- c(127, 125, 3, 127, 50, 50)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
dims <- c(138, 123, 2, 138, 41, 42)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut_male, amgut_female,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
# dissimilarity networks
dims <- c(49, 49, 6, 49, 49, 49)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
#-------------------------------------------------------------------------------
##context("netConstruct with different sample filtering methods (two networks")
set.seed(123456)
fsamp <- c("none", "totalReads", "numbTaxa", "highestFreq")
fsampar <- c(list(totalReads = 1000), list(numbTaxa = 30),list(highestFreq = 100))
dims <- c(289, 289, 288, 100)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1] + dim(testnet$normCounts2)[1], dims[i])
#print(dim(testnet$normCounts1)[1] + dim(testnet$normCounts2)[1])
}
dims1 <- c(146, 127, 128, 100)
dims2 <- c(126, 114, 114, 100)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut_female, amgut_male,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims1[i])
expect_equal(dim(testnet$normCounts2)[1], dims2[i])
#print(dim(testnet$normCounts1)[1])
#print(dim(testnet$normCounts2)[1])
}
dims <- c(289, 270, 228, 13)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "pseudo", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims[i])
#print(dim(testnet$normCounts1)[1])
}
#===============================================================================
# Association / dissimilarity measures
context("netConstruct with different association measures")
set.seed(123456)
measures <- c("pearson", "spearman", "bicor", "sparcc", "propr",
"cclasso", "ccrepe", "gcoda", "spieceasi_gl", "spieceasi_mb", "spring" )
for (i in 1:length(measures)) {
context(measures[i])
measure.tmp <- measures[i]
if (measure.tmp == "spieceasi_gl") {
measurePar <- list(method = "glasso",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.3))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spieceasi_mb") {
measurePar <- list(method = "mb",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.3))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spring") {
measurePar <- list(nlambda = 5, rep.num = 5)
} else if (measure.tmp == "gcoda") {
measurePar <- list(nlambda = 5)
} else {
measurePar <- NULL
}
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = measure.tmp,
measurePar = measurePar,
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
if (i <= 5) {
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
context("test plot.microNetProps")
plot(testprops)
#------------------------------------------------------------------------------
context("netConstruct with different dissimilarity measures")
set.seed(123456)
measures <- c("euclidean", "bray", "kld", "jeffrey", "ckld",
"jsd", "aitchison")
for (i in 1:length(measures)) {
context(measures[i])
testnet1 <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 100),
zeroMethod = "none", normMethod = "none",
measure = measures[i],
sparsMethod = "knn", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet1, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
#===============================================================================
# Zero handling
context("netConstruct with different methods for zero replacement")
set.seed(123456)
zeroMethod <- c("none", "pseudo", "multRepl", "alrEM", "bayesMult")
for (i in 1:length(zeroMethod)) {
context(zeroMethod[i])
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[i], normMethod = "clr",
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(zeroMethod[i], side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
#===============================================================================
# Normalization + zero handling
context("netConstruct with different normalization methods")
set.seed(123456)
normMethod <- c("none", "fractions", "TSS", "CSS", "COM", "rarefy", "clr", "mclr")
zeroMethod <- c("none", "pseudo", "multRepl")
for (i in 1:length(normMethod)) {
for (z in 1:length(zeroMethod)) {
context(paste0("normMethod: ", normMethod[i], "; zeroMethod: ", zeroMethod[z]))
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[z],
normMethod = normMethod[i],
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(paste0(normMethod[i], ", ", zeroMethod[z]), side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
#===============================================================================
# Sparsification
context("netConstruct with different sparsification methods")
set.seed(123456)
sparsMethod <- c("none", "t-test", "bootstrap", "threshold", "softThreshold")
for (i in 1:length(sparsMethod)) {
context(sparsMethod[i])
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 100),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson", adjust = "adaptBH",
sparsMethod = sparsMethod[i], thresh = 0.3,
softThreshPower = c(8,10),
dissFunc = "signed", nboot = 1000,
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(sparsMethod[i], side = 3, cex = 1.5)
if (i != 3) {
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
#===============================================================================
# SoftThreshPower
context("netConstruct with given softThreshPower")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "softThreshold",
softThreshType = "signed",
softThreshPower = c(5,3),
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
stefpeschel/NetCoMi documentation built on Feb. 4, 2024, 8:20 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.
set.seed(123456)
data("amgut1.filt")
data("amgut2.filt.phy")
context("netConstruct with phyloseq object as input")
library(phyloseq)
testnet <- netConstruct(amgut2.filt.phy,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext("phyloseq", side = 3, cex = 1.5)
#===============================================================================
# Filtering
context("netConstruct with different taxa filtering methods (single network)")
set.seed(123456)
ftax <- c("none", "totalReads", "relFreq", "numbSamp", "highestVar", "highestFreq")
ftaxpar <- c(list(totalReads = 1000), list(relFreq = 0.05),
list(numbSamp = 10), list(highestVar = 50), list(highestFreq = 50))
dims <- c(127, 125, 3, 127, 50, 50)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
#-------------------------------------------------------------------------------
context("netConstruct with different sample filtering methods (single network")
set.seed(123456)
fsamp <- c("none", "totalReads", "numbTaxa", "highestFreq")
fsampar <- c(list(totalReads = 1000), list(numbTaxa = 50),list(highestFreq = 50))
dims <- c(289, 289, 282, 50)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims[i])
}
#===============================================================================
# Association / dissimilarity measures
context("netConstruct with different association measures")
set.seed(123456)
measures <- c("pearson", "spearman", "bicor", "sparcc", "cclasso", "ccrepe",
"propr","gcoda", "spieceasi_gl", "spieceasi_mb", "spring" )
for (i in 1:length(measures)) {
context(measures[i])
measure.tmp <- measures[i]
if (measure.tmp == "spieceasi_gl") {
measurePar <- list(method = "glasso",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.2))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spieceasi_mb") {
measurePar <- list(method = "mb",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.2))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spring") {
measurePar <- list(nlambda = 5, rep.num = 5)
} else if (measure.tmp == "gcoda") {
measurePar <- list(nlambda = 5)
} else {
measurePar <- NULL
}
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = measure.tmp,
measurePar = measurePar,
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
}
context("test plot.microNetProps")
plot(testprops)
context("netConstruct with association matrix as input")
set.seed(123456)
testnet1 <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testthat::expect_that(netConstruct(testnet1$assoEst1, dataType = "correlation",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)$assoMat1,
testthat::equals(testnet1$assoMat1))
testthat::expect_that(netConstruct(testnet1$assoEst1, dataType = "correlation",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)$adjaMat1,
testthat::equals(testnet1$adjaMat1))
testnet1 <- netConstruct(amgut1.filt,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 30),
zeroMethod = "none", normMethod = "none",
measure = "ckld",
sparsMethod = "knn", seed = 20190101)
testthat::expect_that(netConstruct(testnet1$dissEst1, dataType = "dissimilarity",
sparsMethod = "knn", seed = 20190101)$dissMat1,
testthat::equals(testnet1$dissMat1))
testthat::expect_that(netConstruct(testnet1$dissEst1, dataType = "dissimilarity",
sparsMethod = "knn", seed = 20190101)$adjaMat1,
testthat::equals(testnet1$adjaMat1))
#-------------------------------------------------------------------------------
context("netConstruct with different dissimilarity measures")
set.seed(123456)
measures <- c("euclidean", "bray", "kld", "jeffrey", "ckld",
"jsd", "aitchison")
for (i in 1:length(measures)) {
context(measures[i])
testnet1 <- netConstruct(amgut1.filt,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 30),
zeroMethod = "none", normMethod = "none",
measure = measures[i],
sparsMethod = "knn", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet1, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
}
context("test plot.microNetProps")
plot(testprops)
#===============================================================================
# Zero handling
context("netConstruct with different methods for zero replacement")
set.seed(123456)
zeroMethod <- c("none", "pseudo", "pseudoZO", "multRepl", "alrEM", "bayesMult")
for (i in 1:length(zeroMethod)) {
context(zeroMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[i], normMethod = "clr",
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(zeroMethod[i], side = 3, cex = 1.5)
}
#===============================================================================
# Normalization + zero handling
context("netConstruct with different normalization methods")
set.seed(123456)
normMethod <- c("none", "fractions", "TSS", "CSS", "COM", "rarefy", "clr", "mclr")
zeroMethod <- c("none", "pseudo", "multRepl")
for (i in 1:length(normMethod)) {
for (z in 1:length(zeroMethod)) {
# context(paste0("normMethod: ", normMethod[i], "; zeroMethod: ", zeroMethod[z]))
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[z],
normMethod = normMethod[i],
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(paste0(normMethod[i], ", ", zeroMethod[z]), side = 3, cex = 1.5)
}
}
#===============================================================================
# Sparsification
context("netConstruct with different sparsification methods")
set.seed(123456)
sparsMethod <- c("none", "t-test", "bootstrap",
"threshold", "softThreshold")
for (i in 1:length(sparsMethod)) {
context(sparsMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 100),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson", adjust = "adaptBH",
sparsMethod = sparsMethod[i],
thresh = 0.3, softThreshPower = 10,
dissFunc = "signed", nboot = 5, cores = 1,
seed = 20190101, logFile = NULL,
verbose = FALSE)
if (sparsMethod[i] != "bootstrap") {
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(sparsMethod[i], side = 3, cex = 1.5)
}
}
#===============================================================================
# Multiple testing adjustment
context("netConstruct with different adjustment methods")
set.seed(123456)
adjustMethod <- c("lfdr", "holm", "BH", "BY", rep("adaptBH", 5))
trueNullMethod <- c(rep("convest", 5), "lfdr", "mean", "hist", "farco")
for (i in 1:length(adjustMethod)) {
context(adjustMethod[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "t-test",
adjust = adjustMethod[i],
lfdrThresh = 0.2,
trueNullMethod = trueNullMethod[i],
dissFunc = "signed", nboot = 1000,
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(adjustMethod[i], side = 3, cex = 1.5)
}
#===============================================================================
# Soft-thresholding
context("netConstruct with different values for softThreshType")
set.seed(123456)
softThreshType <- c("signed", "unsigned", "signed hybrid")
for (i in 1:length(softThreshType)) {
context(softThreshType[i])
if (i == 1) {
softThreshPower <- 5
} else {
softThreshPower <- NULL
}
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "softThreshold",
softThreshType = softThreshType[i],
softThreshPower = softThreshPower,
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(softThreshType[i], side = 3, cex = 1.5)
}
#===============================================================================
# K-nearest neighbor sparsification
context("netConstruct with knn sparsification")
set.seed(123456)
knnMutual <- c(TRUE, FALSE)
for (i in 1:length(knnMutual)) {
context(knnMutual[i])
testnet <- netConstruct(amgut1.filt,
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 50),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "knn",
knnMutual = knnMutual[i],
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(knnMutual[i], side = 3, cex = 1.5)
}
#===============================================================================
# Dissimilarity function
context("netConstruct with differenct values for dissFunc")
set.seed(123456)
dissFunc <- c("signed", "unsigned", "signedPos", "TOMdiss")
for (i in 1:length(dissFunc)) {
context(dissFunc[i])
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = dissFunc[i],
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(dissFunc[i], side = 3, cex = 1.5)
}
#-------------------------------------------------------------------------------
# 'dissFunc' is a function
testfunc <- function(x) {
xvec <- x[lower.tri(x)]
dissvec <- sqrt(0.5 * (1-xvec))
dissMat <- x
dissMat[lower.tri(dissMat)] <- dissvec
dissMat[upper.tri(dissMat)] <- t(dissMat)[upper.tri(t(dissMat))]
dissMat[x == 0] <- Inf
diag(dissMat) <- 0
return(dissMat)
}
context("'dissFunc' is a function")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = testfunc,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext("dissFunc as function", side = 3, cex = 1.5)
#===============================================================================
# Similarity function
testfunc <- function(x, power) {
1/(1 + x^power)
}
context("simFunc")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
dissFunc = "signed",
simFunc = testfunc,
simFuncPar = list(2),
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext("simFunc", side = 3, cex = 1.5)
#===============================================================================
# scaleDiss
context("scaleDiss")
scaleDiss <- c(TRUE, FALSE)
for (i in 1:length(scaleDiss)) {
context(scaleDiss[i])
testnet <- netConstruct(amgut1.filt,
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 50),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "knn",
dissFunc = "signed",
scaleDiss = scaleDiss[i],
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(scaleDiss[i], side = 3, cex = 1.5)
}
#===============================================================================
# Unweighted network
context("unweighted network")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold",
thresh = 0.3,
weighted = FALSE,
seed = 20190101)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
summary(testprops)
plot(testprops)
mtext("weighted", side = 3, cex = 1.5)
################################################################################
### two networks
################################################################################
set.seed(123456)
groups_diss <- sample(0:1, ncol(amgut1.filt), replace = TRUE)
groups_asso <- sample(0:1, nrow(amgut1.filt), replace = TRUE)
amgut_male <- phyloseq::subset_samples(amgut2.filt.phy, SEX == "male")
amgut_female <- phyloseq::subset_samples(amgut2.filt.phy, SEX == "female")
context("netConstruct with group vector (association)")
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with data and data2 (association)")
testnet <- netConstruct(amgut_male, amgut_female,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with group vector (dissimilarity)")
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
context("netConstruct with data and data2 (dissimilarity)")
amgut_diss1 <- amgut1.filt[, groups_diss == 0]
amgut_diss2 <- amgut1.filt[, groups_diss == 1]
testnet <- netConstruct(amgut_diss1, data2 = amgut_diss2,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
#===============================================================================
# Matched-group designs
context("netConstruct with matched-group design")
set.seed(123456)
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:280, ],
matchDesign = c(1,1),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:281, ],
matchDesign = c(1,1),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
testnet <- netConstruct(amgut1.filt[1:95, ], amgut1.filt[96:285, ],
matchDesign = c(1,2),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt[1:140, ], amgut1.filt[141:280, ],
matchDesign = c(1,2),
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
#===============================================================================
# jointPrepro
context("netConstruct with jointPrepro")
set.seed(123456)
testnet <- netConstruct(amgut1.filt, group = groups_asso,
jointPrepro = FALSE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testnet <- netConstruct(amgut_male, amgut_female,
jointPrepro = TRUE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_error(
testnet <- netConstruct(amgut1.filt, group = groups_diss,
jointPrepro = TRUE,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101))
#===============================================================================
# Filtering
##context("netConstruct with different taxa filtering methods (two networks)")
set.seed(123456)
ftax <- c("none", "totalReads", "relFreq", "numbSamp", "highestVar", "highestFreq")
ftaxpar <- c(list(totalReads = 1000), list(relFreq = 0.05),
list(numbSamp = 10), list(highestVar = 50), list(highestFreq = 50))
dims <- c(127, 125, 3, 127, 50, 50)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
dims <- c(138, 123, 2, 138, 41, 42)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut_male, amgut_female,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
# dissimilarity networks
dims <- c(49, 49, 6, 49, 49, 49)
for (i in 1:length(ftax)) {
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = ftax[i],
filtTaxPar = ftaxpar[i-1],
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[2], dims[i])
}
#-------------------------------------------------------------------------------
##context("netConstruct with different sample filtering methods (two networks")
set.seed(123456)
fsamp <- c("none", "totalReads", "numbTaxa", "highestFreq")
fsampar <- c(list(totalReads = 1000), list(numbTaxa = 30),list(highestFreq = 100))
dims <- c(289, 289, 288, 100)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1] + dim(testnet$normCounts2)[1], dims[i])
#print(dim(testnet$normCounts1)[1] + dim(testnet$normCounts2)[1])
}
dims1 <- c(146, 127, 128, 100)
dims2 <- c(126, 114, 114, 100)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut_female, amgut_male,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims1[i])
expect_equal(dim(testnet$normCounts2)[1], dims2[i])
#print(dim(testnet$normCounts1)[1])
#print(dim(testnet$normCounts2)[1])
}
dims <- c(289, 270, 228, 13)
for (i in 1:length(fsamp)) {
testnet <- netConstruct(amgut1.filt, group = groups_diss,
filtSamp = fsamp[i],
filtSampPar = fsampar[i-1],
zeroMethod = "pseudo", normMethod = "none",
measure = "aitchison",
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
expect_equal(dim(testnet$normCounts1)[1], dims[i])
#print(dim(testnet$normCounts1)[1])
}
#===============================================================================
# Association / dissimilarity measures
context("netConstruct with different association measures")
set.seed(123456)
measures <- c("pearson", "spearman", "bicor", "sparcc", "propr",
"cclasso", "ccrepe", "gcoda", "spieceasi_gl", "spieceasi_mb", "spring" )
for (i in 1:length(measures)) {
context(measures[i])
measure.tmp <- measures[i]
if (measure.tmp == "spieceasi_gl") {
measurePar <- list(method = "glasso",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.3))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spieceasi_mb") {
measurePar <- list(method = "mb",
nlambda=5,
pulsar.params = list(rep.num=5,
thresh = 0.3))
measure.tmp <- "spieceasi"
} else if (measure.tmp == "spring") {
measurePar <- list(nlambda = 5, rep.num = 5)
} else if (measure.tmp == "gcoda") {
measurePar <- list(nlambda = 5)
} else {
measurePar <- NULL
}
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = measure.tmp,
measurePar = measurePar,
sparsMethod = "threshold", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
if (i <= 5) {
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
context("test plot.microNetProps")
plot(testprops)
#------------------------------------------------------------------------------
context("netConstruct with different dissimilarity measures")
set.seed(123456)
measures <- c("euclidean", "bray", "kld", "jeffrey", "ckld",
"jsd", "aitchison")
for (i in 1:length(measures)) {
context(measures[i])
testnet1 <- netConstruct(amgut1.filt, group = groups_diss,
filtTax = "totalReads",
filtTaxPar = list(totalReads = 1000),
filtSamp = "highestFreq",
filtSampPar = list(highestFreq = 100),
zeroMethod = "none", normMethod = "none",
measure = measures[i],
sparsMethod = "knn", thresh = 0.3,
seed = 20190101)
testprops<- netAnalyze(testnet1, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(measures[i], side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
#===============================================================================
# Zero handling
context("netConstruct with different methods for zero replacement")
set.seed(123456)
zeroMethod <- c("none", "pseudo", "multRepl", "alrEM", "bayesMult")
for (i in 1:length(zeroMethod)) {
context(zeroMethod[i])
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[i], normMethod = "clr",
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(zeroMethod[i], side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
#===============================================================================
# Normalization + zero handling
context("netConstruct with different normalization methods")
set.seed(123456)
normMethod <- c("none", "fractions", "TSS", "CSS", "COM", "rarefy", "clr", "mclr")
zeroMethod <- c("none", "pseudo", "multRepl")
for (i in 1:length(normMethod)) {
for (z in 1:length(zeroMethod)) {
context(paste0("normMethod: ", normMethod[i], "; zeroMethod: ", zeroMethod[z]))
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 20),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = zeroMethod[z],
normMethod = normMethod[i],
measure = "pearson", adjust = "BY",
sparsMethod = "threshold", thresh = 0.3,
dissFunc = "signed", nboot = 20, cores = 4,
seed = 20190101)
testprops<- netAnalyze(testnet, clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector")
plot(testprops)
mtext(paste0(normMethod[i], ", ", zeroMethod[z]), side = 3, cex = 1.5)
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
#===============================================================================
# Sparsification
context("netConstruct with different sparsification methods")
set.seed(123456)
sparsMethod <- c("none", "t-test", "bootstrap", "threshold", "softThreshold")
for (i in 1:length(sparsMethod)) {
context(sparsMethod[i])
testnet <- netConstruct(amgut1.filt, group = groups_asso,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 100),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson", adjust = "adaptBH",
sparsMethod = sparsMethod[i], thresh = 0.3,
softThreshPower = c(8,10),
dissFunc = "signed", nboot = 1000,
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
mtext(sparsMethod[i], side = 3, cex = 1.5)
if (i != 3) {
netcomp_asso <- netCompare(testprops, permTest = TRUE, nPerm = 4, cores = 1L)
}
}
#===============================================================================
# SoftThreshPower
context("netConstruct with given softThreshPower")
set.seed(123456)
testnet <- netConstruct(amgut1.filt,
filtTax = "highestVar",
filtTaxPar = list(highestVar = 50),
filtSamp = "totalReads",
filtSampPar = list(totalReads = 1000),
zeroMethod = "none", normMethod = "none",
measure = "pearson",
sparsMethod = "softThreshold",
softThreshType = "signed",
softThreshPower = c(5,3),
dissFunc = "signed",
seed = 20190101, logFile = NULL)
testprops<- netAnalyze(testnet,
clustMethod = "cluster_fast_greedy",
hubPar = "eigenvector", normDeg = FALSE,
hubQuant = 0.95, lnormFit = FALSE)
plot(testprops)
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.