data-raw/script_timeOmics.simdata.R

In abodein/timeOmics: Time-Course Multi-Omics data integration

library(tidyverse)
library(lmms)

# RAW DATA
c1 <- c(0, 0.5,1,1.1,1.2,1.8,2.5,5,9)
c3 <-  c(-2,4, 8, 6,4.5,4,3.9, 3, 1)
c2 <- -c1
c4 <- -c3

list(c1,c2,c3,c4)


c1.0 <-  c1
c1.1 <-  c1*1.5
c1.2 <- (c1-0.3)*0.3
c1.3 <- (c1 +0.5)*0.8
c1.4 <- (c1-1)*1.1

c2.0 <-  c2
c2.1 <-  c2*1.5
c2.2 <- (c2-0.3)*0.3
c2.3 <- (c2 +0.5)*0.8
c2.4 <- (c2-1)*1.1

c3.0 <-  c3
c3.1 <-  c3*1.5
c3.2 <- (c3-0.3)*0.3
c3.3 <- (c3 +0.5)*0.8
c3.4 <- (c3-1)*1.1

c4.0 <-  c4
c4.1 <-  c4*1.5
c4.2 <- (c4-0.3)*0.3
c4.3 <- (c4 +0.5)*0.8
c4.4 <- (c4-1)*1.4

# noise 
c0 <- c(0,0.1,0.05,0,0,0.1,0,0.05,0.1) +1
sd(c0)/mean(c0)

data <- list(c1.0,c1.1,c1.2,c1.3,c1.4,c2.0,c2.1,c2.2,c2.3,c2.4,c3.0,c3.1,c3.2,c3.3,c3.4,c4.0,c4.1,c4.2,c4.3,c4.4, c0)
names(data) <- c("c1.0", "c1.1", "c1.2", "c1.3", "c1.4",
                 "c2.0", "c2.1", "c2.2", "c2.3", "c2.4",
                 "c3.0", "c3.1", "c3.2", "c3.3", "c3.4",
                 "c4.0", "c4.1", "c4.2", "c4.3", "c4.4",
                 "c0")
raw.data <- as.data.frame(data)setwd("~")
load("~/Downloads/lmms.out2.rda")
data.gather <- raw.data %>% rownames_to_column("time") %>%
    mutate(time = as.numeric(time)) %>%
    gather(sample, value, -time)

# SIM DATA
sd <- 0.3
N_Ind <- 5
set.seed(123)

tmp <- data.gather
for(ind in 1:N_Ind){
    vect <- vector(length = nrow(tmp), mode = "numeric")
    for(x in 1:length(vect)){
        vect[x] <- rnorm(1, mean = tmp$value[x], sd = sd)
    }
    name.c <- names(tmp)
    tmp <- data.frame(tmp, vect)
    colnames(tmp) <- c(name.c, LETTERS[ind])
}

sim.data <- tmp %>% dplyr::select(-c(value)) %>%
    gather(ind, value, -c(sample, time))%>%
    mutate(ind = c(paste0(ind, "_", time))) %>% dplyr::select(-time) %>%
    spread(ind, value) %>% column_to_rownames("sample") %>% t

# modelled data
time <- rownames(sim.data) %>% str_split("_") %>% map_chr(~.x[2]) %>% as.numeric()
sampleID <- rownames(sim.data)
lmms.out <- lmmSpline(data = sim.data, time = time, sampleID = sampleID, keepModels = TRUE)

# build new s4
#setClass('lmms',slots=c(basis="character", knots="numeric",errorMolecules="character"))
setClass("lmmspline",slots= c(predSpline="data.frame", modelsUsed="numeric",models="list",derivative='logical', basis="character", knots="numeric",errorMolecules="character"))

lmms.out2 <- new("lmmspline", 
                 predSpline = lmms.out@predSpline,
                 modelsUsed = lmms.out@modelsUsed,
                 models = lmms.out@models,
                 derivative = lmms.out@derivative, 
                 basis = lmms.out@basis, 
                 knots = lmms.out@knots, 
                 errorMolecules = lmms.out@errorMolecules )

save(lmms.out2, file = "~/Downloads/lmms.out2.rda")

modelled.data <-  as.data.frame(t(lmms.out2@predSpline))

detach("package:lmms", unload=TRUE)

timeOmics.simdata <- list(rawdata = raw.data, sim = sim.data,
                          modelled = modelled.data[,-c0], 
                          lmms.output = lmms.out2,
                          time = time)

library(lmms)
# Y same as data but increase noise
sd <- 0.5
N_Ind <- 4
set.seed(123)

tmp <- data.gather %>% filter(time %in% c(1,2,3,5,7,9))
for(ind in 1:N_Ind){
    vect <- vector(length = nrow(tmp), mode = "numeric")
    for(x in 1:length(vect)){
        vect[x] <- rnorm(1, mean = tmp$value[x], sd = sd)
    }
    name.c <- names(tmp)
    tmp <- data.frame(tmp, vect)
    colnames(tmp) <- c(name.c, LETTERS[ind])
}

Y <- tmp %>% dplyr::select(-c(value)) %>%
    gather(ind, value, -c(sample, time))%>%
    mutate(ind = c(paste0(ind, "_", time))) %>% dplyr::select(-time) %>%
    spread(ind, value) %>% column_to_rownames("sample") %>% t

time.Y <- rownames(Y) %>% str_split("_") %>% map_chr(~.x[2]) %>% as.numeric()
sampleID.Y <- rownames(Y)
lmms.Y <- lmmSpline(data = Y, time = time.Y, sampleID = sampleID.Y, keepModels = TRUE, 
                          timePredict = 1:9)
modelled.Y <- lmms.Y@predSpline %>% t %>% as.data.frame()
colnames(modelled.Y) <- paste0("Y_", seq_along(colnames(modelled.Y)))

timeOmics.simdata[["Y"]] <- modelled.Y

# Z
# Y same as data but increase noise
sd <- 1
N_Ind <- 4
set.seed(123)

tmp <- data.gather %>% filter(time %in% c(1,3,4,5,8,9))
for(ind in 1:N_Ind){
    vect <- vector(length = nrow(tmp), mode = "numeric")
    for(x in 1:length(vect)){
        vect[x] <- rnorm(1, mean = tmp$value[x], sd = sd)
    }
    name.c <- names(tmp)
    tmp <- data.frame(tmp, vect)
    colnames(tmp) <- c(name.c, LETTERS[ind])
}

Z <- tmp %>% dplyr::select(-c(value)) %>%
    gather(ind, value, -c(sample, time))%>%
    mutate(ind = c(paste0(ind, "_", time))) %>% dplyr::select(-time) %>%
    spread(ind, value) %>% column_to_rownames("sample") %>% t

time.Z <- rownames(Z) %>% str_split("_") %>% map_chr(~.x[2]) %>% as.numeric()
sampleID.Z <- rownames(Z)
lmms.Z <- lmms::lmmSpline(data = Z, time = time.Z, sampleID = sampleID.Z, keepModels = TRUE, 
                          timePredict = 1:9)
modelled.Z <- lmms.Z@predSpline %>% t %>% as.data.frame()
colnames(modelled.Z) <- paste0("Z_", seq_along(colnames(modelled.Z)))

timeOmics.simdata[["Z"]] <- modelled.Z

usethis::use_data(timeOmics.simdata, overwrite = TRUE)
#save(timeOmics.simdata, file = "./data/timeOmics.simdata.rda", compress = "gzip", ascii = FALSE)