Last update: 24-10-16
# version: 22.03.24
knitr::opts_chunk$set(warning = FALSE, message = FALSE, results = FALSE)
# dirs
## Sebastian @Linux
dir_moa <- "/home/isar/Dropbox/scn_moa/"
## Sebastian @MAC
#dir_moa <- "/Users/home_sh/Library/CloudStorage/Dropbox/scn_moa"
# data, revised 2024.01.09
# exp
load(file = paste0(dir_moa, "/data/final/", "gnp.pro.RData"))
load(file = paste0(dir_moa, "/data/final/", "gnp.mrn.RData"))
load(file = paste0(dir_moa, "/data/final/", "gnp.mic.RData"))
#load(file = paste0(dir_moa, "/data/final/", "scn.pro.RData")) # discarded prom project
# results
load(file = paste0(dir_moa, "/data/final/", "dex.all.RData"))
load(file = paste0(dir_moa, "/data/final/preps/timeomics/", "spca.timeOmics.RData"))
#load(file = paste0(dir_moa, "/data/final/", "timeomics.singles.RData"))
#load(file = paste0(dir_moa, "/data/final/", "res.wgcna.RData"))
## packages
source(paste0(dir_moa, "/scripts/", "scn_moa_fct.R"))
#source(paste0(dir_moa, "/scripts/FINAL/plotter/", "feature_plots.R"))
# load packagea
sapply(packages_general, require, character.only = TRUE)
sapply(packages_enrichment, require, character.only = TRUE)
library(timeOmics)
library("viridis")
library(gt)mrn
prep
# copy time omics data
meds.mrn <- spca.mrn
# timeOmics plot and extract the data
plot_dat0.mrn <- plotLong(meds.mrn)
# change time to stage for traj plots
plot_dat0.mrn %<>% mutate(time = recode(time,
'1' = 'MB',
'2' = 'PM',
'3' = 'MC',
'4' = 'MM',
'5' = 'B',
'6' = 'S',
'7' = 'PMN'))
# Calculate mean expression values
plot_dat.mrn <-
plot_dat0.mrn %>%
group_by(time, comp, contribution) %>% # could I use the cluster instead? probably yes
mutate(mean_exp = mean(value)) %>%
ungroup()
## calculate the centrality
# make wide
plot_dat_wide.mrn <- plot_dat0.mrn %>%
#mutate(time = paste0("Time", time)) %>%
pivot_wider(names_from = time, values_from = value)
# collect the PCs and contributions
combo.mrn <- expand.grid(
unique(plot_dat.mrn$comp %>% as.character()) %>% sort,
unique(plot_dat.mrn$contribution %>% as.character())
)
# get centrality
out_centrality.mrn <- vector("list", nrow(combo.mrn))
for(i in 1:nrow(combo.mrn)){
comp_ind <- combo.mrn[i, 1]
contr_ind <- combo.mrn[i, 2]
dat <-
plot_dat_wide.mrn %>%
as.data.frame() %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
`rownames<-`(.$molecule) %>%
dplyr::select(MB:PMN) %>%
as.matrix()
dat <- dat %>%
as.data.frame() %>%
mutate(centrality = 1 - pchisq(rowSums(scale(dat)^2), df = 6))
out_centrality.mrn[[i]] <- dat
}
# collect the plots
outPlots.mrn <- vector("list", nrow(combo.mrn))
for(i in 1:nrow(combo.mrn)){
comp_ind <- combo.mrn[i, 1]
contr_ind <- combo.mrn[i, 2]
centrality <- out_centrality.mrn[[i]]$centrality
cols <- rep(centrality, 7)
outPlots.mrn[[i]] <-
plot_dat.mrn %>%
mutate(time = factor(time, levels = levels_mrn)) %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
mutate(cols = cols) %>%
arrange(cols) %>%
ggplot() +
geom_line(aes(x = time, y = value, group = cols, colour = cols)) +
scale_colour_viridis_c(option = "D") +
theme_light(base_size = 18)+
geom_line(aes(x = time, y = mean_exp, group = cols), colour = "red", linewidth = 1.5) +
theme(axis.text.x=element_text(angle=30, hjust=1))
}
ggarrange(plotlist = outPlots.mrn)
# save plots
plotCollection.mrn <- outPlots.mrnplots
# mrn down
p.cat.mrn_down <- plotCollection.mrn[[1]]
p.cat.mrn_down <- p.cat.mrn_down + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mrn_down
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_down", ".png"),
plot = p.cat.mrn_down,
device = png,
width = 3,
height = 3)
# mrn up
p.cat.mrn_up <- plotCollection.mrn[[5]]
p.cat.mrn_up <- p.cat.mrn_up + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mrn_up
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_up", ".png"),
plot = p.cat.mrn_up,
device = png,
width = 3,
height = 3)
# mrn up down
p.cat.mrn_upDn <- plotCollection.mrn[[6]]
p.cat.mrn_upDn <- p.cat.mrn_upDn + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.mrn_upDn
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_upDn", ".png"),
plot = p.cat.mrn_upDn,
device = png,
width = 3,
height = 3)
# mrn up down
p.cat.mrn_dnUp <- plotCollection.mrn[[2]]
p.cat.mrn_dnUp <- p.cat.mrn_dnUp + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mrn_dnUp
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_dnUp", ".png"),
plot = p.cat.mrn_dnUp,
device = png,
width = 3,
height = 3)
# mrn up down up
p.cat.mrn_upDnUp <- plotCollection.mrn[[7]]
p.cat.mrn_upDnUp <- p.cat.mrn_upDnUp + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.mrn_upDnUp
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_upDnUp", ".png"),
plot = p.cat.mrn_upDnUp,
device = png,
width = 3,
height = 3)
# mrn up down
p.cat.mrn_dnUpDn <- plotCollection.mrn[[3]]
p.cat.mrn_dnUpDn <- p.cat.mrn_dnUpDn + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mrn_dnUpDn
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mrn_dnUpDn", ".png"),
plot = p.cat.mrn_dnUpDn,
device = png,
width = 3,
height = 3)pro
Last 3 stages combined ## prep
# copy of the original time omics data
meds.pro.cFin <- spca.pro
# plots and extraction of the data
plot_dat0.pro.cFin <- plotLong(meds.pro.cFin)
# change time to stage for plotting
plot_dat0.pro.cFin %<>% mutate(time = recode(time,
'1' = 'MB',
'2' = 'PM',
'3' = 'MC',
'4' = 'Mature'))
# Calculate mean expression
plot_dat.pro.cFin <-
plot_dat0.pro.cFin %>%
group_by(time, comp, contribution) %>%
mutate(mean_exp = mean(value)) %>%
ungroup()
## calculate the centrality
# make wide
plot_dat_wide.pro.cFin <- plot_dat0.pro.cFin %>%
#mutate(time = paste0("Time", time)) %>%
pivot_wider(names_from = time, values_from = value)
# combine
combo.pro.cFin <- expand.grid(
unique(plot_dat.pro.cFin$comp %>% as.character()) %>% sort,
unique(plot_dat.pro.cFin$contribution %>% as.character())
)
# get centrality
out_centrality.pro.cFin <- vector("list", nrow(combo.pro.cFin))
for(i in 1:nrow(combo.pro.cFin)){
comp_ind <- combo.pro.cFin[i, 1]
contr_ind <- combo.pro.cFin[i, 2]
dat <-
plot_dat_wide.pro.cFin %>%
as.data.frame() %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
`rownames<-`(.$molecule) %>%
dplyr::select(MB:Mature) %>%
as.matrix()
dat <- dat %>%
as.data.frame() %>%
mutate(centrality = 1 - pchisq(rowSums(scale(dat)^2), df = 3)) # degrees of freedom: n stages - 1
out_centrality.pro.cFin[[i]] <- dat
}
# collect the plots
outPlots.pro.cFin <- vector("list", nrow(combo.pro.cFin))
for(i in 1:nrow(combo.pro.cFin)){
comp_ind <- combo.pro.cFin[i, 1]
contr_ind <- combo.pro.cFin[i, 2]
centrality <- out_centrality.pro.cFin[[i]]$centrality
cols <- rep(centrality, 4) # correct to change this number to the number of cols???
outPlots.pro.cFin[[i]] <-
plot_dat.pro.cFin %>%
mutate(time = factor(time, levels = c('MB', 'PM', 'MC', 'Mature'))) %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
mutate(cols = cols) %>%
arrange(cols) %>%
ggplot() +
geom_line(aes(x = time, y = value, group = cols, colour = cols)) +
scale_colour_viridis_c(option = "D") +
theme_light(base_size = 18)+
geom_line(aes(x = time, y = mean_exp, group = cols), colour = "red", linewidth = 1.5) +
theme(axis.text.x=element_text(angle=30, hjust=1))
}
ggarrange(plotlist = outPlots.pro.cFin)
# save plots
plotCollection.pro.cFin <- outPlots.pro.cFinplots
# up
p.cat.pro_up <- plotCollection.pro.cFin[[1]]
p.cat.pro_up <- p.cat.pro_up + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.pro_up
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_up", ".png"),
plot = p.cat.pro_up,
device = png,
width = 3,
height = 3)
# up down
p.cat.pro_upDn <- plotCollection.pro.cFin[[2]]
p.cat.pro_upDn <- p.cat.pro_upDn + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.pro_upDn
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_upDn", ".png"),
plot = p.cat.pro_upDn,
device = png,
width = 3,
height = 2.8)
# up down up
p.cat.pro_upDnUp <- plotCollection.pro.cFin[[3]]
p.cat.pro_upDnUp <- p.cat.pro_upDnUp + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.pro_upDnUp
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_upDnUp", ".png"),
plot = p.cat.pro_upDnUp,
device = png,
width = 3,
height = 3)
# down
p.cat.pro_down <- plotCollection.pro.cFin[[4]]
p.cat.pro_down <- p.cat.pro_down + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.pro_down
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_down", ".png"),
plot = p.cat.pro_down,
device = png,
width = 3,
height = 3)
# down up
p.cat.pro_dnUp <- plotCollection.pro.cFin[[5]]
p.cat.pro_dnUp <- p.cat.pro_dnUp + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.pro_dnUp
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_dnUp", ".png"),
plot = p.cat.pro_dnUp,
device = png,
width = 3,
height = 3)
# down up down
p.cat.pro_dnUpDn <- plotCollection.pro.cFin[[6]]
p.cat.pro_dnUpDn <- p.cat.pro_dnUpDn + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.pro_dnUpDn
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.pro_dnUpDn", ".png"),
plot = p.cat.pro_dnUpDn,
device = png,
width = 3,
height = 3)mic
prep
# copy time omics data
meds.mic <- spca.mic
# timeOmics plot and extract the data
plot_dat0.mic <- plotLong(meds.mic)
table(plot_dat0.mic$time)
# change time to stage for traj plots
plot_dat0.mic %<>% mutate(time = recode(time,
'1' = 'MB',
'2' = 'MC',
'3' = 'MM',
'4' = 'B',
'5' = 'S',
'6' = 'PMN'))
# Calculate mean expression values
plot_dat.mic <-
plot_dat0.mic %>%
group_by(time, comp, contribution) %>% # could I use the cluster instead? probably yes
mutate(mean_exp = mean(value)) %>%
ungroup()
## calculate the centrality
# make wide
plot_dat_wide.mic <- plot_dat0.mic %>%
#mutate(time = paste0("Time", time)) %>%
pivot_wider(names_from = time, values_from = value)
# collect the PCs and contributions
combo.mic <- expand.grid(
unique(plot_dat.mic$comp %>% as.character()) %>% sort,
unique(plot_dat.mic$contribution %>% as.character())
)
# get centrality
out_centrality.mic <- vector("list", nrow(combo.mic))
for(i in 1:nrow(combo.mic)){
comp_ind <- combo.mic[i, 1]
contr_ind <- combo.mic[i, 2]
dat <-
plot_dat_wide.mic %>%
as.data.frame() %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
`rownames<-`(.$molecule) %>%
dplyr::select(MB:PMN) %>%
as.matrix()
dat <- dat %>%
as.data.frame() %>%
mutate(centrality = 1 - pchisq(rowSums(scale(dat)^2), df = 5))
out_centrality.mic[[i]] <- dat
}
# collect the plots
outPlots.mic <- vector("list", nrow(combo.mic))
for(i in 1:nrow(combo.mic)){
comp_ind <- combo.mic[i, 1]
contr_ind <- combo.mic[i, 2]
centrality <- out_centrality.mic[[i]]$centrality
cols <- rep(centrality, 6) # adapt to stage n
outPlots.mic[[i]] <-
plot_dat.mic %>%
mutate(time = factor(time, levels = levels_mic)) %>%
dplyr::filter(comp == comp_ind & contribution == contr_ind) %>%
mutate(cols = cols) %>%
arrange(cols) %>%
ggplot() +
geom_line(aes(x = time, y = value, group = cols, colour = cols)) +
scale_colour_viridis_c(option = "D") +
theme_light(base_size = 18)+
geom_line(aes(x = time, y = mean_exp, group = cols), colour = "red", linewidth = 1.5) +
theme(axis.text.x=element_text(angle=30, hjust=1))
}
ggarrange(plotlist = outPlots.mic)
# save plots
plotCollection.mic <- outPlots.micplots
# mic down
p.cat.mic_down <- plotCollection.mic[[1]]
p.cat.mic_down <- p.cat.mic_down + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mic_down
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mic_down", ".png"),
plot = p.cat.mic_down,
device = png,
width = 3,
height = 3)
# mic down up
p.cat.mic_dnUp <- plotCollection.mic[[2]]
p.cat.mic_dnUp <- p.cat.mic_dnUp + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mic_dnUp
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mic_dnUp", ".png"),
plot = p.cat.mic_dnUp,
device = png,
width = 3,
height = 3)
# mic up
p.cat.mic_up <- plotCollection.mic[[4]]
p.cat.mic_up <- p.cat.mic_up + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank())
p.cat.mic_up
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mic_up", ".png"),
plot = p.cat.mic_up,
device = png,
width = 3,
height = 3)
# mic up down
p.cat.mic_upDn <- plotCollection.mic[[5]]
p.cat.mic_upDn <- p.cat.mic_upDn + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.mic_upDn
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mic_upDn", ".png"),
plot = p.cat.mic_upDn,
device = png,
width = 3,
height = 2.8)
# mic up down
p.cat.mic_upDn.late <- plotCollection.mic[[6]]
p.cat.mic_upDn.late <- p.cat.mic_upDn.late + theme(legend.position="none", axis.title.x = element_blank(), axis.title.y = element_blank(), axis.text.x=element_blank())
p.cat.mic_upDn.late
ggsave(
filename = paste0(dir_moa, "/figures/fig2/", "p.cat.mic_upDn.late", ".png"),
plot = p.cat.mic_upDn.late,
device = png,
width = 3,
height = 3)extract features & numbers
mrn
# work copy
dat.mrn4feat <- plot_dat.mrn
# rename clusters: from -1 via 1 to 4
plotLong(meds.mrn)
table(dat.mrn4feat$cluster)
dat.mrn4feat %<>%
mutate(traj = recode(cluster,
'1' = 'up',
'-1' = 'dn',
'-2' = 'dnUp',
'2' = 'upDn',
'-3' = 'dnUpDn',
'3' = 'upDnUp',
'-4' = 'dnUpDnUp',
'4' = 'upDnUpDn'))
# create numers table
numbers.mrn.raw <- dat.mrn4feat %>% distinct(molecule, .keep_all = T) %>%
pull(traj) %>% table() %>% as.data.frame()
names(numbers.mrn.raw) <- c('traj', 'n.raw')
## now try to limit the features per cluster to the ones with high centrality (= 55)
centrality.mrn <- bind_rows(out_centrality.mrn, .id = "cluster")
centrality.mrn$ENSG <- rownames(centrality.mrn)
table(centrality.mrn$cluster) # these are 1:8
ggarrange(plotlist = outPlots.mrn)
centrality.mrn %<>%
mutate(traj = recode(cluster,
'1' = 'dn',
'2' = 'dnUp',
'3' = 'dnUpDn',
'4' = 'dnUpDnUp',
'5' = 'up',
'6' = 'upDn',
'7' = 'upDnup',
'8' = 'upDnUpDn'))
centrality.mrn.90 <- centrality.mrn %>% select(ENSG, traj, centrality) %>%
group_by(traj) %>%
arrange(traj, desc(centrality)) %>%
filter(centrality > quantile(centrality, .10))
numbers.mrn.90 <- table(centrality.mrn.90$traj) %>% as.data.frame()
names(numbers.mrn.90) <- c('traj', 'n.90')
numbers.mrn <- left_join(numbers.mrn.raw, numbers.mrn.90, by = c('traj'))
add.percentage <- function(df, num.col, baseNum, newColName){
df[[newColName]] <- round(df[[num.col]] / baseNum * 100, 0)
df[[newColName]] <- paste0("(", df[[newColName]], "%)")
return(df)
}
tot.mrn <- ncol(gnp.mrn$exp_mrn_tpm_vsn)
numbers.mrn <- add.percentage(numbers.mrn, 'n.raw', tot.mrn, 'prc.raw') # its debatable of we should use the raw number or the input?
numbers.mrn <- add.percentage(numbers.mrn, 'n.90', tot.mrn, 'prc.90')
numbers.mrn %>% select(traj, n.raw, prc.raw, n.90, prc.90) %>% gt()
tab.mrn <- numbers.mrn %>% select(traj, n= n.90, prc = prc.90) %>% gt()
tab.mrn %>% gtsave(filename = paste0(dir_moa, "/figures/fig2/", "tab.mrn", '.png'))
361+313pro
# work copy
dat.pro4feat <- plot_dat.pro.cFin
# rename clusters: from -1 via 1 to 4
plotLong(meds.pro.cFin)
table(dat.pro4feat$cluster)
dat.pro4feat %<>%
mutate(traj = recode(cluster,
'1' = 'up',
'-1' = 'dn',
'-2' = 'dnUp',
'2' = 'upDn',
'-3' = 'dnUpDn',
'3' = 'upDnUp'))
# create numers table
numbers.pro.raw <- dat.pro4feat %>% distinct(molecule, .keep_all = T) %>%
pull(traj) %>% table() %>% as.data.frame()
names(numbers.pro.raw) <- c('traj', 'n.raw')
## now try to limit the features per cluster to the ones with high centrality (= 55)
centrality.pro <- bind_rows(out_centrality.pro.cFin, .id = "cluster")
centrality.pro$ENSG <- rownames(centrality.pro)
table(centrality.pro$cluster) # these are 1:8
ggarrange(plotlist = outPlots.pro.cFin)
centrality.pro %<>%
mutate(traj = recode(cluster,
'1' = 'up',
'2' = 'upDn',
'3' = 'upDnUp',
'4' = 'dn',
'5' = 'dnUp',
'6' = 'dnUpDn'))
centrality.pro.90 <- centrality.pro %>% select(ENSG, traj, centrality) %>%
group_by(traj) %>%
arrange(traj, desc(centrality)) %>%
filter(centrality > quantile(centrality, .10))
numbers.pro.90 <- table(centrality.pro.90$traj) %>% as.data.frame()
names(numbers.pro.90) <- c('traj', 'n.90')
numbers.pro <- left_join(numbers.pro.raw, numbers.pro.90, by = c('traj'))
add.percentage <- function(df, num.col, baseNum, newColName){
df[[newColName]] <- round(df[[num.col]] / baseNum * 100, 0)
df[[newColName]] <- paste0("(", df[[newColName]], "%)")
return(df)
}
tot.pro <- ncol(gnp.pro$exp_pro_vsn)
numbers.pro <- add.percentage(numbers.pro, 'n.raw', tot.pro, 'prc.raw') # its debatable of we should use the raw number or the input?
numbers.pro <- add.percentage(numbers.pro, 'n.90', tot.pro, 'prc.90')
numbers.pro %>% select(traj, n.raw, prc.raw, n.90, prc.90) %>% gt()
tab.pro <- numbers.pro %>% select(traj, n= n.90, prc = prc.90) %>% gt()
tab.pro %>% gtsave(filename = paste0(dir_moa, "/figures/fig2/", "tab.pro", '.png'))
329+261mic
# work copy
dat.mic4feat <- plot_dat.mic
# rename clusters: from -1 via 1 to 4
plotLong(meds.mic)
table(dat.mic4feat$cluster)
dat.mic4feat %<>%
mutate(traj = recode(cluster,
'1' = 'up',
'-1' = 'dn',
'-2' = 'dnUp',
'2' = 'upDn',
'-3' = 'NA',
'3' = 'upDn late'))
# create numers table
numbers.mic.raw <- dat.mic4feat %>% distinct(molecule, .keep_all = T) %>%
pull(traj) %>% table() %>% as.data.frame()
names(numbers.mic.raw) <- c('traj', 'n.raw')
## now try to limit the features per cluster to the ones with high centrality (= 55)
centrality.mic <- bind_rows(out_centrality.mic, .id = "cluster")
centrality.mic$ENSG <- rownames(centrality.mic)
table(centrality.mic$cluster) # these are 1:8
ggarrange(plotlist = outPlots.mic)
centrality.mic %<>%
mutate(traj = recode(cluster,
'1' = 'dn',
'2' = 'dnUp',
'3' = 'NA',
'4' = 'up',
'5' = 'upDn',
'6' = 'upDn late'))
centrality.mic.90 <- centrality.mic %>% select(ENSG, traj, centrality) %>%
group_by(traj) %>%
arrange(traj, desc(centrality)) %>%
filter(centrality > quantile(centrality, .10))
numbers.mic.90 <- table(centrality.mic.90$traj) %>% as.data.frame()
names(numbers.mic.90) <- c('traj', 'n.90')
numbers.mic <- left_join(numbers.mic.raw, numbers.mic.90, by = c('traj'))
add.percentage <- function(df, num.col, baseNum, newColName){
df[[newColName]] <- round(df[[num.col]] / baseNum * 100, 0)
df[[newColName]] <- paste0("(", df[[newColName]], "%)")
return(df)
}
tot.mic <- ncol(gnp.mic$exp_mic_vsn)
numbers.mic <- add.percentage(numbers.mic, 'n.raw', tot.mic, 'prc.raw') # its debatable of we should use the raw number or the input?
numbers.mic <- add.percentage(numbers.mic, 'n.90', tot.mic, 'prc.90')
numbers.mic %>% select(traj, n.raw, prc.raw, n.90, prc.90) %>% gt()
tab.mic <- numbers.mic %>% select(traj, n= n.90, prc = prc.90) %>% gt()
tab.mic %>% gtsave(filename = paste0(dir_moa, "/figures/fig2/", "tab.mic", '.png'))
21+12export
# export
save(
centrality.pro,
centrality.mic,
centrality.mrn,
file = paste0(dir_moa, "/data/final/preps/timeomics/", "centralities.RData"))