Integrative network pharmacology, transcriptomics, and molecular docking identify candidate Centella asiatica constituents and targets in neurodegenerative diseases.

Code ↔ Paper

The paper beside its authors' code: matches between them have not been computed for this paper yet.

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

Markdown · 1,285 lines · 37 KB · CC-BY-4.0

# Reproducibility README
This file contains an anonymized and consolidated R workflow for the analyses performed in the manuscript:
**Identification of Potential Bioactive Constituents of *Centella asiatica* for Neurodegenerative Diseases Using Network Pharmacology and Molecular Docking**
The workflow was consolidated from the original analysis scripts used for compound filtering, target prediction integration, disease-gene integration, PCA, differential-expression analysis, GO/KEGG enrichment, network-file construction, machine-learning feature selection, hub-gene expression validation, and ROC analysis.
All local paths and project-specific filenames have been replaced with generic placeholders. Before running the code, replace the placeholder input filenames with the corresponding files in your repository.
---
## Recommended repository structure
```text
project_root/
├── input/
│ ├── compound_screening/
│ ├── target_prediction/
│ ├── disease_genes/
│ ├── expression_matrices/
│ ├── candidate_genes/
│ └── docking_outputs/
├── output/
│ ├── compound_screening/
│ ├── target_prediction/
│ ├── disease_genes/
│ ├── PCA/
│ ├── DEG/
│ ├── enrichment/
│ ├── networks/
│ ├── machine_learning/
│ ├── expression_validation/
│ └── ROC/
└── README.md
```
---
## Input-file conventions
The workflow assumes the following generic input formats.
### 1. Expression matrix
Rows are genes and columns are samples.
```text
Gene Sample01_Control Sample02_Control Sample03_Disease
GENE1 5.21 5.34 7.18
GENE2 8.02 7.95 6.11
```
Sample names should end with `_Control` or `_Disease`.
### 2. Gene list
One gene symbol per line.
```text
GENE1
GENE2
GENE3
```
### 3. SwissADME output
The SwissADME table should contain columns such as molecular weight, hydrogen-bond donors, hydrogen-bond acceptors, rotatable bonds, LogP values, gastrointestinal absorption, and BBB permeability.
### 4. SwissTargetPrediction output
Each compound should have a target-prediction CSV file. The target-gene column and probability column should be checked before running the script.
---
## R workflow
```r
# ============================================================
# 0. Setup
# ============================================================
set.seed(12345)
# Install packages manually if needed.
# install.packages(c("dplyr", "ggplot2", "ggpubr", "pheatmap",
# "ggrepel", "ggvenn", "glmnet", "randomForest",
# "e1071", "caret", "pROC", "data.table",
# "tidyverse", "RColorBrewer"))
# if (!requireNamespace("BiocManager", quietly = TRUE)) install.packages("BiocManager")
# BiocManager::install(c("limma", "sva", "clusterProfiler", "org.Hs.eg.db", "enrichplot", "DESeq2"))
suppressPackageStartupMessages({
library(limma)
library(sva)
library(dplyr)
library(ggplot2)
library(ggpubr)
library(pheatmap)
library(ggrepel)
library(ggvenn)
library(clusterProfiler)
library(org.Hs.eg.db)
library(enrichplot)
library(glmnet)
library(randomForest)
library(e1071)
library(caret)
library(pROC)
library(data.table)
library(tidyverse)
})
dir.create("output", showWarnings = FALSE, recursive = TRUE)
# ============================================================
# Helper functions
# ============================================================
read_expression_matrix_simple <- function(file) {
expr <- read.table(
file,
header = TRUE,
sep = "\t",
check.names = FALSE,
row.names = 1,
comment.char = "",
quote = ""
)
expr <- as.matrix(expr)
mode(expr) <- "numeric"
expr <- avereps(expr)
return(expr)
}
extract_group <- function(sample_names) {
group <- sub("^.*_", "", sample_names)
group <- ifelse(
grepl("^Control$", group, ignore.case = TRUE),
"Control",
ifelse(grepl("^Disease$", group, ignore.case = TRUE), "Disease", NA)
)
if (any(is.na(group))) {
stop(
"Sample names must end with '_Control' or '_Disease'. Problematic samples: ",
paste(sample_names[is.na(group)], collapse = ", ")
)
}
factor(group, levels = c("Control", "Disease"))
}
write_vector <- function(x, file) {
write.table(
unique(x),
file = file,
sep = "\t",
quote = FALSE,
row.names = FALSE,
col.names = FALSE
)
}
ensure_dir <- function(path) {
dir.create(path, showWarnings = FALSE, recursive = TRUE)
}
# ============================================================
# 1. Candidate-compound filtering using SwissADME outputs
# ============================================================
filter_swissadme_compounds <- function(
swissadme_file,
output_file,
mw_max = 500,
hbd_max = 5,
hba_max = 10,
rotatable_max = 10,
logp_min = -2,
logp_max = 5,
require_high_gi = TRUE,
require_bbb = TRUE
) {
adme <- read.csv(
swissadme_file,
header = TRUE,
sep = ",",
check.names = FALSE
)
filtered <- subset(
adme,
MW < mw_max &
`#H-bond donors` < hbd_max &
`#H-bond acceptors` < hba_max &
`#Rotatable bonds` < rotatable_max &
iLOGP > logp_min & iLOGP < logp_max &
XLOGP3 > logp_min & XLOGP3 < logp_max &
`Consensus Log P` > logp_min & `Consensus Log P` < logp_max
)
if (require_high_gi) {
filtered <- subset(filtered, `GI absorption` == "High")
}
if (require_bbb) {
filtered <- subset(filtered, `BBB permeant` == "Yes")
}
write.csv(filtered, output_file, row.names = FALSE)
return(filtered)
}
# Example:
# filtered_compounds <- filter_swissadme_compounds(
# swissadme_file = "input/compound_screening/swissadme_results.csv",
# output_file = "output/compound_screening/filtered_compounds.csv"
# )
# ============================================================
# 2. SwissTargetPrediction target integration
# ============================================================
integrate_predicted_targets <- function(
target_prediction_dir,
filtered_compound_file,
output_file,
probability_cutoff = 0.05,
compound_id_col = 1,
compound_name_col = "Ingredient name",
target_gene_col = 2,
probability_col = "Probability*"
) {
compound_info <- read.csv(
filtered_compound_file,
header = TRUE,
sep = ",",
check.names = FALSE,
row.names = compound_id_col
)
csv_files <- list.files(
target_prediction_dir,
pattern = "\\.csv$",
full.names = TRUE
)
out <- data.frame()
for (file in csv_files) {
target_table <- read.csv(
file,
header = TRUE,
sep = ",",
check.names = FALSE
)
if (!(probability_col %in% colnames(target_table))) next
target_table <- subset(target_table, target_table[[probability_col]] > probability_cutoff)
if (nrow(target_table) == 0) next
genes <- unlist(strsplit(as.vector(target_table[[target_gene_col]]), " "))
genes <- trimws(genes)
genes <- unique(genes[genes != ""])
compound_id <- tools::file_path_sans_ext(basename(file))
compound_name <- compound_info[compound_id, compound_name_col]
if (length(genes) > 0) {
out <- rbind(
out,
data.frame(
Compound_ID = compound_id,
Compound = compound_name,
Gene = genes,
stringsAsFactors = FALSE
)
)
}
}
out <- unique(out)
write.table(out, output_file, sep = "\t", quote = FALSE, row.names = FALSE)
return(out)
}
# Example:
# predicted_targets <- integrate_predicted_targets(
# target_prediction_dir = "input/target_prediction/",
# filtered_compound_file = "output/compound_screening/filtered_compounds.csv",
# output_file = "output/target_prediction/compound_target_pairs.tsv"
# )
# ============================================================
# 3. Disease-gene integration and Venn plot
# ============================================================
integrate_disease_genes <- function(
disease_gene_dir,
output_union_file,
output_venn_pdf
) {
files <- list.files(disease_gene_dir, pattern = "\\.txt$", full.names = TRUE)
gene_list <- list()
for (file in files) {
tab <- read.table(file, header = FALSE, sep = "\t", check.names = FALSE)
genes <- unlist(strsplit(as.vector(tab[, 1]), " "))
genes <- trimws(genes)
genes <- unique(genes[genes != ""])
source_name <- tools::file_path_sans_ext(basename(file))
gene_list[[source_name]] <- genes
}
pdf(output_venn_pdf, width = 6, height = 6)
print(
ggvenn(
gene_list,
show_percentage = TRUE,
stroke_color = "white",
stroke_size = 0.5,
fill_color = c("#E41A1C", "#1E90FF", "#FF8C00", "#31A354FF"),
set_name_color = c("#E41A1C", "#1E90FF", "#FF8C00", "#31A354FF"),
set_name_size = 6,
text_size = 4.5
)
)
dev.off()
union_genes <- Reduce(union, gene_list)
write_vector(union_genes, output_union_file)
return(union_genes)
}
# Example:
# disease_genes <- integrate_disease_genes(
# disease_gene_dir = "input/disease_genes/",
# output_union_file = "output/disease_genes/disease_gene_union.tsv",
# output_venn_pdf = "output/disease_genes/disease_gene_venn.pdf"
# )
# ============================================================
# 4. PCA before and after batch correction
# ============================================================
plot_pca <- function(
expression_file,
output_pdf,
title
) {
expr <- read_expression_matrix_simple(expression_file)
data <- t(expr)
dataset <- gsub("(.*?)\\_.*", "\\1", rownames(data))
pca <- prcomp(data, scale. = FALSE)
pred <- predict(pca)
pca_df <- data.frame(
PC1 = pred[, 1],
PC2 = pred[, 2],
Dataset = dataset
)
pdf(output_pdf, width = 5.5, height = 4.25)
print(
ggscatter(
data = pca_df,
x = "PC1",
y = "PC2",
color = "Dataset",
shape = "Dataset",
ellipse = TRUE,
ellipse.type = "norm",
ellipse.border.remove = FALSE,
ellipse.alpha = 0.1,
size = 2,
main = title,
legend = "right"
) +
theme(
plot.margin = unit(rep(1.5, 4), "lines"),
plot.title = element_text(hjust = 0.5)
)
)
dev.off()
}
# Example:
# plot_pca("input/expression_matrices/discovery_pre_batch_correction.tsv",
# "output/PCA/PCA_before_batch_correction.pdf",
# "Before batch correction")
# plot_pca("input/expression_matrices/discovery_post_batch_correction.tsv",
# "output/PCA/PCA_after_batch_correction.pdf",
# "After batch correction")
# ============================================================
# 5. Differential-expression analysis for microarray data
# ============================================================
run_limma_deg <- function(
expression_file,
output_dir,
logfc_cutoff = 0.5,
p_cutoff = 0.05,
adj_p_cutoff = 0.05
) {
ensure_dir(output_dir)
expr <- read_expression_matrix_simple(expression_file)
group <- extract_group(colnames(expr))
expr <- expr[, order(group)]
group <- extract_group(colnames(expr))
design <- model.matrix(~0 + group)
colnames(design) <- levels(group)
fit <- lmFit(expr, design)
contrast_matrix <- makeContrasts(Disease - Control, levels = design)
fit2 <- contrasts.fit(fit, contrast_matrix)
fit2 <- eBayes(fit2)
deg_all <- topTable(fit2, adjust.method = "BH", number = Inf)
deg_all$Gene <- rownames(deg_all)
deg_nominal <- deg_all %>%
filter(P.Value < p_cutoff & abs(logFC) > logfc_cutoff)
deg_fdr <- deg_all %>%
filter(adj.P.Val < adj_p_cutoff & abs(logFC) > logfc_cutoff)
write.table(deg_all, file.path(output_dir, "all_DEG_results.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
write.table(deg_nominal, file.path(output_dir, "DEGs_nominalP.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
write.table(deg_fdr, file.path(output_dir, "DEGs_FDR.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
if (nrow(deg_nominal) > 0) {
write_vector(deg_nominal$Gene, file.path(output_dir, "DEG_gene_list.tsv"))
}
return(list(all = deg_all, nominal = deg_nominal, fdr = deg_fdr))
}
# Example:
# deg_results <- run_limma_deg(
# expression_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# output_dir = "output/DEG/"
# )
# ============================================================
# 6. GO enrichment analysis
# ============================================================
run_go_enrichment <- function(
gene_file,
output_dir,
pvalue_cutoff = 0.05,
padj_cutoff = 0.05
) {
ensure_dir(output_dir)
genes <- read.table(gene_file, header = FALSE, sep = "\t", check.names = FALSE)[, 1]
genes <- unique(genes)
entrez <- mget(genes, org.Hs.egSYMBOL2EG, ifnotfound = NA)
entrez <- as.character(entrez)
entrez <- entrez[entrez != "NA"]
ego <- enrichGO(
gene = entrez,
OrgDb = org.Hs.eg.db,
pvalueCutoff = 1,
qvalueCutoff = 1,
ont = "all",
readable = TRUE
)
go_df <- as.data.frame(ego)
go_sig <- go_df %>%
filter(pvalue < pvalue_cutoff & p.adjust < padj_cutoff)
write.table(go_sig, file.path(output_dir, "GO_enrichment_results.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
pdf(file.path(output_dir, "GO_barplot.pdf"), width = 8.5, height = 7)
print(
barplot(
ego,
drop = TRUE,
showCategory = 10,
label_format = 100,
split = "ONTOLOGY",
color = "p.adjust"
) +
facet_grid(ONTOLOGY ~ ., scale = "free")
)
dev.off()
return(go_sig)
}
# Example:
# go_results <- run_go_enrichment(
# gene_file = "input/candidate_genes/overlapping_genes.tsv",
# output_dir = "output/enrichment/GO/"
# )
# ============================================================
# 7. KEGG enrichment analysis
# ============================================================
run_kegg_enrichment <- function(
gene_file,
output_dir,
pvalue_cutoff = 0.05,
padj_cutoff = 0.05,
show_n = 20
) {
ensure_dir(output_dir)
genes <- read.table(gene_file, header = FALSE, sep = "\t", check.names = FALSE)[, 1]
genes <- unique(genes)
entrez <- mget(genes, org.Hs.egSYMBOL2EG, ifnotfound = NA)
entrez <- as.character(entrez)
mapping <- data.frame(Gene = genes, EntrezID = entrez, stringsAsFactors = FALSE)
entrez <- entrez[entrez != "NA"]
ekegg <- enrichKEGG(
gene = entrez,
organism = "hsa",
pvalueCutoff = 1,
qvalueCutoff = 1
)
kegg_df <- as.data.frame(ekegg)
if (nrow(kegg_df) > 0) {
kegg_df$geneID <- as.character(sapply(kegg_df$geneID, function(x) {
paste(mapping$Gene[match(strsplit(x, "/")[[1]], as.character(mapping$EntrezID))],
collapse = "/")
}))
}
kegg_sig <- kegg_df %>%
filter(pvalue < pvalue_cutoff & p.adjust < padj_cutoff) %>%
filter(category != "Human Diseases") %>%
na.omit()
write.table(kegg_sig, file.path(output_dir, "KEGG_enrichment_results.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
if (nrow(kegg_sig) > 0) {
show_kegg <- kegg_sig[seq_len(min(show_n, nrow(kegg_sig))), ]
show_kegg$Pathway <- factor(show_kegg$Description, levels = rev(show_kegg$Description))
p <- ggplot(show_kegg, aes(x = Count, y = Pathway, fill = p.adjust)) +
geom_bar(stat = "identity", width = 0.75) +
scale_fill_distiller(palette = "Spectral", direction = 1) +
labs(x = "Gene count", y = "", title = "Enriched KEGG pathways") +
theme_bw() +
theme(
plot.title = element_text(size = 12, hjust = 0.5, face = "bold"),
axis.title = element_text(size = 11),
axis.text = element_text(size = 10),
legend.title = element_text(size = 11),
legend.text = element_text(size = 10)
)
pdf(file.path(output_dir, "KEGG_barplot.pdf"), width = 7, height = 5.5)
print(p)
dev.off()
}
return(kegg_sig)
}
# Example:
# kegg_results <- run_kegg_enrichment(
# gene_file = "input/candidate_genes/overlapping_genes.tsv",
# output_dir = "output/enrichment/KEGG/"
# )
# ============================================================
# 8. Network-file construction for Cytoscape
# ============================================================
build_compound_target_pathway_network <- function(
compound_target_file,
kegg_file,
output_dir,
top_kegg_n = 10
) {
ensure_dir(output_dir)
compound_target <- read.table(
compound_target_file,
header = TRUE,
sep = "\t",
check.names = FALSE,
comment.char = "",
quote = ""
)
kegg <- read.table(
kegg_file,
header = TRUE,
sep = "\t",
check.names = FALSE,
comment.char = "",
quote = ""
)
kegg_top <- kegg[order(kegg$p.adjust, kegg$pvalue), ]
kegg_top <- head(kegg_top, top_kegg_n)
drug_name <- unique(compound_target$Drug)[1]
disease_name <- unique(compound_target$Disease)[1]
compound_list <- unique(compound_target$Compound)
gene_list <- unique(compound_target$Gene)
pathway_list <- unique(kegg_top$Description)
network <- data.frame()
for (compound in compound_list) {
network <- rbind(
network,
data.frame(Node1 = drug_name, Node2 = compound, Interaction = "Drug-Compound")
)
genes <- unique(compound_target$Gene[compound_target$Compound == compound])
genes <- genes[!is.na(genes) & genes != ""]
if (length(genes) > 0) {
network <- rbind(
network,
data.frame(Node1 = compound, Node2 = genes, Interaction = "Compound-Target")
)
}
}
for (i in seq_len(nrow(kegg_top))) {
pathway <- kegg_top$Description[i]
network <- rbind(
network,
data.frame(Node1 = disease_name, Node2 = pathway, Interaction = "Disease-Pathway")
)
genes <- unlist(strsplit(as.vector(kegg_top$geneID[i]), "/"))
genes <- trimws(unique(genes))
genes <- intersect(genes, gene_list)
if (length(genes) > 0) {
network <- rbind(
network,
data.frame(Node1 = pathway, Node2 = genes, Interaction = "Pathway-Target")
)
}
}
network <- unique(network)
all_nodes <- unique(c(network$Node1, network$Node2))
node_type <- ifelse(
all_nodes == drug_name, "Drug",
ifelse(
all_nodes == disease_name, "Disease",
ifelse(
all_nodes %in% compound_list, "Compound",
ifelse(
all_nodes %in% pathway_list, "Pathway",
ifelse(all_nodes %in% gene_list, "Target", "Other")
)
)
)
)
nodes <- data.frame(Node = all_nodes, Type = node_type)
write.table(network, file.path(output_dir, "network_edges.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
write.table(nodes, file.path(output_dir, "network_nodes.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
write_vector(gene_list, file.path(output_dir, "network_gene_list.tsv"))
write_vector(compound_list, file.path(output_dir, "network_compound_list.tsv"))
write_vector(pathway_list, file.path(output_dir, "network_pathway_list.tsv"))
return(list(edges = network, nodes = nodes))
}
# Example:
# network_files <- build_compound_target_pathway_network(
# compound_target_file = "input/network/compound_target_info.tsv",
# kegg_file = "output/enrichment/KEGG/KEGG_enrichment_results.tsv",
# output_dir = "output/networks/"
# )
# ============================================================
# 9. LASSO feature selection
# ============================================================
run_lasso_feature_selection <- function(
expression_file,
candidate_gene_file,
output_dir,
nfolds = 10
) {
ensure_dir(output_dir)
expr <- read_expression_matrix_simple(expression_file)
candidate_genes <- read.table(candidate_gene_file, header = FALSE, sep = "\t")[, 1]
candidate_genes <- intersect(unique(candidate_genes), rownames(expr))
expr <- expr[candidate_genes, , drop = FALSE]
x <- as.matrix(t(expr))
y <- extract_group(rownames(x))
fit <- glmnet(x, y, family = "binomial", alpha = 1)
cvfit <- cv.glmnet(
x,
y,
family = "binomial",
alpha = 1,
type.measure = "deviance",
nfolds = nfolds
)
pdf(file.path(output_dir, "LASSO_coefficient_path.pdf"), width = 6, height = 5.5)
plot(fit)
dev.off()
pdf(file.path(output_dir, "LASSO_cross_validation.pdf"), width = 6, height = 5.5)
plot(cvfit)
dev.off()
coef_min <- coef(cvfit, s = "lambda.min")
selected <- rownames(coef_min)[which(as.numeric(coef_min) != 0)]
selected <- setdiff(selected, "(Intercept)")
write_vector(selected, file.path(output_dir, "LASSO_selected_genes.tsv"))
return(selected)
}
# Example:
# lasso_genes <- run_lasso_feature_selection(
# expression_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# candidate_gene_file = "input/candidate_genes/overlapping_genes.tsv",
# output_dir = "output/machine_learning/LASSO/"
# )
# ============================================================
# 10. Random-forest feature selection
# ============================================================
run_random_forest_feature_selection <- function(
expression_file,
candidate_gene_file,
output_dir,
repeat_times = 1000,
ntree = 500,
top_n = 10,
frequency_cutoff = 0.5
) {
ensure_dir(output_dir)
expr <- read_expression_matrix_simple(expression_file)
candidate_genes <- read.table(candidate_gene_file, header = FALSE, sep = "\t")[, 1]
candidate_genes <- intersect(unique(candidate_genes), rownames(expr))
expr <- expr[candidate_genes, , drop = FALSE]
data <- as.data.frame(t(expr))
group <- extract_group(rownames(data))
selected_list <- list()
for (i in seq_len(repeat_times)) {
rf <- randomForest(as.factor(group) ~ ., data = data, ntree = ntree)
imp <- importance(rf)
top_genes <- rownames(imp)[order(imp[, 1], decreasing = TRUE)[seq_len(min(top_n, nrow(imp)))]]
selected_list[[i]] <- top_genes
}
gene_frequency <- sort(table(unlist(selected_list)), decreasing = TRUE)
stable_genes <- names(gene_frequency[gene_frequency >= repeat_times * frequency_cutoff])
write.table(
data.frame(Gene = names(gene_frequency), Frequency = as.numeric(gene_frequency)),
file.path(output_dir, "RF_gene_selection_frequency.tsv"),
sep = "\t",
quote = FALSE,
row.names = FALSE
)
write_vector(stable_genes, file.path(output_dir, "RF_stable_genes.tsv"))
return(stable_genes)
}
# Example:
# rf_genes <- run_random_forest_feature_selection(
# expression_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# candidate_gene_file = "input/candidate_genes/overlapping_genes.tsv",
# output_dir = "output/machine_learning/RF/",
# top_n = 10
# )
# ============================================================
# 11. SVM-RFE feature selection
# ============================================================
run_svm_rfe_feature_selection <- function(
expression_file,
candidate_gene_file,
output_dir,
cv_number = 10,
cv_repeats = 5,
acc_ratio = 0.998,
min_gene_n = 4,
max_gene_n = 10
) {
ensure_dir(output_dir)
expr <- read_expression_matrix_simple(expression_file)
candidate_genes <- read.table(candidate_gene_file, header = FALSE, sep = "\t")[, 1]
candidate_genes <- intersect(unique(candidate_genes), rownames(expr))
expr <- expr[candidate_genes, , drop = FALSE]
data <- as.data.frame(t(expr))
group <- extract_group(rownames(data))
data$group <- group
gene_var <- apply(data[, candidate_genes, drop = FALSE], 2, var)
candidate_genes <- names(gene_var[gene_var > 0])
data <- data[, c(candidate_genes, "group")]
x <- data[, candidate_genes, drop = FALSE]
y <- data$group
svm_fit <- svm(x = x, y = y, kernel = "linear", scale = TRUE)
w <- t(svm_fit$coefs) %*% svm_fit$SV
gene_rank <- data.frame(
Gene = colnames(x),
Weight = as.numeric(w),
AbsWeight = abs(as.numeric(w))
)
gene_rank <- gene_rank[order(gene_rank$AbsWeight, decreasing = TRUE), ]
write.table(gene_rank, file.path(output_dir, "SVM_gene_ranking.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
ctrl <- trainControl(
method = "repeatedcv",
number = cv_number,
repeats = cv_repeats,
classProbs = TRUE,
summaryFunction = defaultSummary,
savePredictions = "final"
)
svm_result <- data.frame()
for (i in seq_len(nrow(gene_rank))) {
use_genes <- gene_rank$Gene[seq_len(i)]
tmp_data <- data[, c(use_genes, "group")]
fit <- train(
group ~ .,
data = tmp_data,
method = "svmLinear",
metric = "Accuracy",
trControl = ctrl,
preProcess = c("center", "scale")
)
acc <- max(fit$results$Accuracy, na.rm = TRUE)
svm_result <- rbind(
svm_result,
data.frame(GeneNumber = i, Accuracy = acc, Error = 1 - acc)
)
}
write.table(svm_result, file.path(output_dir, "SVM_RFE_performance.tsv"),
sep = "\t", quote = FALSE, row.names = FALSE)
best_acc <- max(svm_result$Accuracy, na.rm = TRUE)
threshold <- best_acc * acc_ratio
max_gene_n <- min(max_gene_n, nrow(gene_rank))
candidate_n <- svm_result$GeneNumber[
svm_result$Accuracy >= threshold &
svm_result$GeneNumber >= min_gene_n &
svm_result$GeneNumber <= max_gene_n
]
if (length(candidate_n) == 0) {
tmp <- svm_result[
svm_result$GeneNumber >= min_gene_n &
svm_result$GeneNumber <= max_gene_n, ]
best_gene_n <- tmp$GeneNumber[which.max(tmp$Accuracy)]
} else {
best_gene_n <- min(candidate_n)
}
selected <- gene_rank$Gene[seq_len(best_gene_n)]
write_vector(selected, file.path(output_dir, "SVM_RFE_selected_genes.tsv"))
pdf(file.path(output_dir, "SVM_RFE_accuracy.pdf"), width = 7, height = 6)
print(
ggplot(svm_result, aes(x = GeneNumber, y = Accuracy)) +
geom_line(linewidth = 1.2, color = "#50C878") +
geom_point(size = 2.5, color = "#50C878") +
geom_hline(yintercept = threshold, linetype = "dotted", color = "gray40") +
geom_vline(xintercept = best_gene_n, linetype = "dashed", color = "red") +
annotate("text", x = best_gene_n,
y = svm_result$Accuracy[svm_result$GeneNumber == best_gene_n],
label = paste0("n=", best_gene_n),
color = "red",
fontface = "bold",
size = 5,
vjust = -1) +
theme_bw() +
labs(x = "Number of features", y = "Cross-validation accuracy",
title = "SVM-RFE accuracy")
)
dev.off()
return(selected)
}
# Example:
# svm_genes <- run_svm_rfe_feature_selection(
# expression_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# candidate_gene_file = "input/candidate_genes/overlapping_genes.tsv",
# output_dir = "output/machine_learning/SVM_RFE/"
# )
# ============================================================
# 12. Hub-gene expression validation
# ============================================================
plot_hub_gene_expression <- function(
discovery_expr_file,
validation_expr_file,
hub_gene_file,
output_pdf
) {
discovery_expr <- read_expression_matrix_simple(discovery_expr_file)
validation_expr <- read_expression_matrix_simple(validation_expr_file)
hub_genes <- read.table(hub_gene_file, header = FALSE, sep = "\t")[, 1]
hub_genes <- unique(hub_genes)
discovery_expr <- discovery_expr[intersect(hub_genes, rownames(discovery_expr)), , drop = FALSE]
validation_expr <- validation_expr[intersect(hub_genes, rownames(validation_expr)), , drop = FALSE]
reshape_expression <- function(expr, dataset_name) {
df <- as.data.frame(expr)
df$Gene <- rownames(df)
df_long <- df %>%
pivot_longer(-Gene, names_to = "Sample", values_to = "Expression") %>%
mutate(
Group = case_when(
grepl("_Control$", Sample) ~ "Control",
grepl("_Disease$", Sample) ~ "Disease",
TRUE ~ NA_character_
),
Dataset = dataset_name
) %>%
filter(!is.na(Group))
df_long
}
discovery_long <- reshape_expression(discovery_expr, "Discovery")
validation_long <- reshape_expression(validation_expr, "Validation")
all_data <- bind_rows(discovery_long, validation_long)
all_data$Group <- factor(all_data$Group, levels = c("Control", "Disease"))
all_data$Dataset <- factor(all_data$Dataset, levels = c("Discovery", "Validation"))
p <- ggboxplot(
all_data,
x = "Dataset",
y = "Expression",
color = "Group",
fill = "Group",
add = "jitter",
facet.by = "Gene",
short.panel.labs = TRUE,
nrow = 1,
xlab = "",
ylab = "Expression",
legend = "top"
) +
stat_compare_means(
aes(group = Group),
method = "t.test",
label = "p.signif",
label.y.npc = "top",
hide.ns = TRUE
) +
theme_bw(base_size = 12) +
theme(
strip.text = element_text(face = "bold", size = 12),
panel.grid = element_blank(),
axis.text.x = element_text(angle = 45, hjust = 1),
legend.title = element_blank()
)
pdf(output_pdf, width = 12, height = 4.5)
print(p)
dev.off()
return(all_data)
}
# Example:
# hub_expression_data <- plot_hub_gene_expression(
# discovery_expr_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# validation_expr_file = "input/expression_matrices/validation_expression.tsv",
# hub_gene_file = "input/candidate_genes/final_hub_genes.tsv",
# output_pdf = "output/expression_validation/hub_gene_expression.pdf"
# )
# ============================================================
# 13. ROC analysis for single genes and multigene model
# ============================================================
run_roc_analysis <- function(
discovery_expr_file,
validation_expr_file,
hub_gene_file,
output_dir,
k = 5
) {
ensure_dir(output_dir)
discovery_expr <- read_expression_matrix_simple(discovery_expr_file)
validation_expr <- read_expression_matrix_simple(validation_expr_file)
discovery_group <- extract_group(colnames(discovery_expr))
validation_group <- extract_group(colnames(validation_expr))
hub_genes <- read.table(hub_gene_file, header = FALSE, sep = "\t")[, 1]
hub_genes <- unique(hub_genes)
hub_genes <- intersect(hub_genes, rownames(discovery_expr))
hub_genes <- intersect(hub_genes, rownames(validation_expr))
if (length(hub_genes) < 1) {
stop("No hub genes were found in both discovery and validation datasets.")
}
discovery_df <- as.data.frame(t(discovery_expr[hub_genes, , drop = FALSE]))
validation_df <- as.data.frame(t(validation_expr[hub_genes, , drop = FALSE]))
discovery_df$group <- discovery_group
validation_df$group <- validation_group
# Single-gene ROC in discovery and validation datasets
single_roc_summary <- data.frame()
for (dataset_name in c("Discovery", "Validation")) {
if (dataset_name == "Discovery") {
df <- discovery_df
group <- discovery_group
} else {
df <- validation_df
group <- validation_group
}
pdf(file.path(output_dir, paste0("single_gene_ROC_", dataset_name, ".pdf")),
width = 6, height = 5)
auc_text <- c()
colors <- rainbow(length(hub_genes))
for (i in seq_along(hub_genes)) {
gene <- hub_genes[i]
roc_obj <- roc(
response = df$group,
predictor = as.numeric(df[[gene]]),
levels = c("Control", "Disease"),
direction = "auto",
ci = TRUE
)
ci_obj <- ci.auc(roc_obj)
single_roc_summary <- rbind(
single_roc_summary,
data.frame(
Dataset = dataset_name,
Gene = gene,
AUC = as.numeric(auc(roc_obj)),
CI_low = as.numeric(ci_obj[1]),
CI_high = as.numeric(ci_obj[3])
)
)
if (i == 1) {
plot(roc_obj, col = colors[i], lwd = 3, legacy.axes = TRUE,
main = paste0("Single-gene ROC curves in ", tolower(dataset_name), " dataset"))
} else {
plot(roc_obj, col = colors[i], lwd = 3, legacy.axes = TRUE, add = TRUE)
}
auc_text <- c(auc_text, paste0(gene, ", AUC=", sprintf("%.3f", auc(roc_obj))))
}
legend("bottomright", legend = auc_text, col = colors, lwd = 3, bty = "n", cex = 0.8)
dev.off()
}
write.csv(single_roc_summary,
file.path(output_dir, "single_gene_ROC_summary.csv"),
row.names = FALSE)
# Five-fold cross-validation in discovery dataset
x_cv <- discovery_df[, hub_genes, drop = FALSE]
x_cv_scaled <- as.data.frame(scale(x_cv))
colnames(x_cv_scaled) <- make.names(colnames(x_cv_scaled))
cv_df <- x_cv_scaled
cv_df$group <- discovery_df$group
fold_id <- sample(rep(seq_len(k), length.out = nrow(cv_df)))
cv_pred <- rep(NA, nrow(cv_df))
for (fold in seq_len(k)) {
train_idx <- which(fold_id != fold)
test_idx <- which(fold_id == fold)
train_fold <- cv_df[train_idx, , drop = FALSE]
test_fold <- cv_df[test_idx, , drop = FALSE]
formula_cv <- as.formula(
paste("group ~", paste(colnames(x_cv_scaled), collapse = " + "))
)
fit <- glm(formula_cv, data = train_fold, family = binomial(link = "logit"))
cv_pred[test_idx] <- predict(fit, newdata = test_fold, type = "response")
}
roc_cv <- roc(
response = cv_df$group,
predictor = cv_pred,
levels = c("Control", "Disease"),
direction = "auto",
ci = TRUE
)
cv_ci <- ci.auc(roc_cv)
pdf(file.path(output_dir, "multigene_model_ROC_discovery_5foldCV.pdf"),
width = 5.5, height = 5)
plot(roc_cv, print.auc = TRUE, col = "red", legacy.axes = TRUE,
main = "Five-fold CV ROC in discovery dataset", lwd = 3)
text(0.45, 0.35,
paste0("95% CI: ", sprintf("%.3f", cv_ci[1]), "-", sprintf("%.3f", cv_ci[3])),
col = "red")
dev.off()
# Train discovery model and apply to external validation dataset
x_train <- discovery_df[, hub_genes, drop = FALSE]
x_test <- validation_df[, hub_genes, drop = FALSE]
train_mean <- apply(x_train, 2, mean, na.rm = TRUE)
train_sd <- apply(x_train, 2, sd, na.rm = TRUE)
x_train_scaled <- as.data.frame(scale(x_train, center = train_mean, scale = train_sd))
x_test_scaled <- as.data.frame(scale(x_test, center = train_mean, scale = train_sd))
colnames(x_train_scaled) <- make.names(colnames(x_train_scaled))
colnames(x_test_scaled) <- make.names(colnames(x_test_scaled))
model_df <- x_train_scaled
model_df$group <- discovery_df$group
formula_model <- as.formula(
paste("group ~", paste(colnames(x_train_scaled), collapse = " + "))
)
fit_model <- glm(formula_model, data = model_df, family = binomial(link = "logit"))
validation_pred <- predict(fit_model, newdata = x_test_scaled, type = "response")
roc_validation <- roc(
response = validation_df$group,
predictor = validation_pred,
levels = c("Control", "Disease"),
direction = "auto",
ci = TRUE
)
validation_ci <- ci.auc(roc_validation)
pdf(file.path(output_dir, "multigene_model_ROC_external_validation.pdf"),
width = 5.5, height = 5)
plot(roc_validation, print.auc = TRUE, col = "red", legacy.axes = TRUE,
main = "External validation ROC of multigene model", lwd = 3)
text(0.45, 0.35,
paste0("95% CI: ", sprintf("%.3f", validation_ci[1]), "-", sprintf("%.3f", validation_ci[3])),
col = "red")
dev.off()
model_summary <- data.frame(
Model = c("Multigene model, five-fold CV in discovery dataset",
"Multigene model, external validation"),
AUC = c(as.numeric(auc(roc_cv)), as.numeric(auc(roc_validation))),
CI_low = c(as.numeric(cv_ci[1]), as.numeric(validation_ci[1])),
CI_high = c(as.numeric(cv_ci[3]), as.numeric(validation_ci[3]))
)
write.csv(model_summary,
file.path(output_dir, "multigene_ROC_summary.csv"),
row.names = FALSE)
return(list(single_gene = single_roc_summary, multigene = model_summary))
}
# Example:
# roc_results <- run_roc_analysis(
# discovery_expr_file = "input/expression_matrices/discovery_post_batch_correction.tsv",
# validation_expr_file = "input/expression_matrices/validation_expression.tsv",
# hub_gene_file = "input/candidate_genes/final_hub_genes.tsv",
# output_dir = "output/ROC/"
# )
# ============================================================
# 14. Session information
# ============================================================
# Save the R session information for reproducibility.
# writeLines(capture.output(sessionInfo()), "output/sessionInfo.txt")
```
---
## Notes for reuse
1. Replace all placeholder filenames with the files in your repository.
2. Keep input sample names ending with `_Control` or `_Disease`.
3. The microarray differential-expression workflow uses `limma` and reports Benjamini-Hochberg adjusted *P* values.
4. The ROC workflow uses `pROC::roc()` and `pROC::ci.auc()` to calculate AUC values and confidence intervals.
5. External validation datasets are used only for validation and are not used for feature selection or model fitting.
6. The code is intended for reproducibility of the analysis workflow and should be accompanied by processed input matrices, sample metadata, and output tables.
---
## Suggested citation statement for the manuscript repository
```text
All author-generated R code used for data processing, differential-expression analysis, enrichment analysis, machine-learning feature selection, ROC analysis and figure generation is provided in this repository. Local paths and non-public filenames have been replaced with generic placeholders to facilitate reuse.
```

README.md, under CC-BY-4.0 · at the source

Overview

Authors: Yuxi Xie¹, Chong-Teik Lim², Xin-Jieh Lam³, Pike-See Cheah^4,5, King-Hwa Ling^2,5,6,7, Tan Huang⁸

ORCID iDs: Pike-See Cheah, King-Hwa Ling, Tan Huang

Institutions: Zunyi Medical University (China); Universiti Putra Malaysia (Malaysia); University of Toyama (Japan); Universitas Pembangunan Nasional Veteran Jakarta (Indonesia); Universiti Tunku Abdul Rahman Sungai Long Campus (Malaysia); Universiti Tunku Abdul Rahman (Malaysia)

Journal: PloS one, volume 21, issue 7, article e0354882

Dates: received 22 April 2026; accepted 12 July 2026; published online 31 July 2026

Type: Research article · Language: English

License: CC BY

Identifiers: DOI 10.1371/journal.pone.0354882 · PMID 42536688 · PMCID PMC13426974 · OpenAlex W7171985819

Open access: gold, a free copy (OpenAlex)

Status: code verified

Categories: genetics / omics (modality), human (organism), other condition (population), cellular / molecular (subfield)

Methods: Statistics, Smoothing, state filtering, decompositions, Machine learning

MeSH: Centella*, Molecular Docking Simulation*, Network Pharmacology*, Neurodegenerative Diseases*, Transcriptome*, Triterpenes*, Gene Expression Profiling, Humans (* major topic)

Topic: Medicinal Plants and Neuroprotection (Complementary and alternative medicine, Medicine), according to OpenAlex

Citations: not cited yet (Europe PMC); 68 references in the paper

Abstract

Background: Neurodegenerative diseases, including Alzheimer’s disease (AD), Parkinson’s disease (PD), and Huntington’s disease (HD), are progressive disorders with limited therapeutic options. Centella asiatica (C. asiatica), a medicinal and edible plant, has been reported to exert neuroprotective and anti-neuroinflammatory properties. Yet, the mechanisms underlying its effects against neurodegenerative diseases remain largely unclear.

Methods: We employed an integrative strategy combining network pharmacology, transcriptomic analyses, machine learning and molecular docking to prioritize disease-associated molecular networks and candidate compound–target relationships in AD, PD and HD.

Results: Sixteen candidate constituents of C. asiatica met the predefined drug-likeness, gastrointestinal absorption and blood–brain barrier permeability criteria, yielding 370 unique predicted targets. Disease-gene mining identified 983 AD-associated genes, 1,103 PD-associated genes, and 3,316 HD-associated genes. Integration of compound targets, disease-associated genes, and transcriptomic profiles prioritized five hub genes in PD (CCKAR, MAPK8, PSEN2, SLC6A3, and TH), four in AD (APP, PGK1, PIK3CA, and TTR), and four in HD (CHRND, HSP90AA1, PRKCQ, and TH). Enrichment analyses highlighted disease-relevant processes involving neurotransmitter signalling, cAMP and calcium pathways, MAPK-related responses and inflammatory regulation. ROC analyses provided additional support for the discriminatory performance of the prioritized genes in independent datasets, whereas molecular docking identified favourable predicted Vina docking scores and structurally plausible interactions between selected compounds and hub targets.

Conclusion: This integrative computational analysis prioritizes candidate C. asiatica constituents, putative disease-associated targets, and molecular pathways in AD, PD, and HD. The findings provide a foundation for subsequent biochemical, cellular, and in vivo validation.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above.

Zenodo 20779507

License: CC-BY-4.0

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Size: 1 file

Software Heritage: not checked

Found in: “Data Availability”

Not found: README, license file, CITATION.cff, environment file, tests, continuous integration, documentation

Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)

27 September 2026: the link answers (HTTP 200)

1 file

README.md, Text, 1,285 lines

At the source: doi.org/10.5281/zenodo.20779507

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
0 scripts, each with its path and the digest of its content;
no match between paragraphs and code yet;
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

geo:GSE20163, at NCBI GEO; found in the text, “Transcriptomic datasets acquisition and…”

Data Availability

No new experimental raw data were generated in this study. The raw transcriptomic datasets analyzed in this study are publicly available from the Gene Expression Omnibus under accession numbers GSE20163, GSE20164, GSE26927, GSE7621, GSE5281, GSE36980, GSE118553, GSE64810 and GSE33000. Processed gene-expression matrices, differential-expression results, and anonymized analysis code generated in this study have been deposited in Zenodo and are available at https://doi.org/10.5281/zenodo.20779507.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 6 authors, 8 MeSH terms, 68 references.

Cite

This paper

Xie, Y., Lim, C.-T., Lam, X.-J., Cheah, P.-S., Ling, K.-H., & Huang, T. (2026). Integrative network pharmacology, transcriptomics, and molecular docking identify candidate Centella asiatica constituents and targets in neurodegenerative diseases. PloS one, 21(7), e0354882. https://doi.org/10.1371/journal.pone.0354882

BibTeX

@article{xie2026integrative, author = {Xie, Yuxi and Lim, Chong-Teik and Lam, Xin-Jieh and Cheah, Pike-See and Ling, King-Hwa and Huang, Tan}, title = {{Integrative network pharmacology, transcriptomics, and molecular docking identify candidate Centella asiatica constituents and targets in neurodegenerative diseases}}, journal = {PloS one}, year = {2026}, month = jul, volume = {21}, number = {7}, pages = {e0354882}, publisher = {PLOS}, issn = {1932-6203}, doi = {10.1371/journal.pone.0354882}, url = {https://doi.org/10.1371/journal.pone.0354882}, pmid = {42536688}, pmcid = {PMC13426974} }

RIS

TY - JOUR AU - Xie, Yuxi AU - Lim, Chong-Teik AU - Lam, Xin-Jieh AU - Cheah, Pike-See AU - Ling, King-Hwa AU - Huang, Tan TI - Integrative network pharmacology, transcriptomics, and molecular docking identify candidate Centella asiatica constituents and targets in neurodegenerative diseases T2 - PloS one J2 - PLoS One PY - 2026 DA - 2026/07/31 VL - 21 IS - 7 SP - e0354882 SN - 1932-6203 PB - PLOS DO - 10.1371/journal.pone.0354882 UR - https://doi.org/10.1371/journal.pone.0354882 LA - en ER -

CSL-JSON

{ "id": "10.1371/journal.pone.0354882", "type": "article-journal", "title": "Integrative network pharmacology, transcriptomics, and molecular docking identify candidate Centella asiatica constituents and targets in neurodegenerative diseases", "container-title": "PloS one", "author": [ { "family": "Xie", "given": "Yuxi" }, { "family": "Lim", "given": "Chong-Teik" }, { "family": "Lam", "given": "Xin-Jieh" }, { "family": "Cheah", "given": "Pike-See" }, { "family": "Ling", "given": "King-Hwa" }, { "family": "Huang", "given": "Tan" } ], "container-title-short": "PLoS One", "volume": "21", "issue": "7", "page": "e0354882", "DOI": "10.1371/journal.pone.0354882", "PMID": "42536688", "PMCID": "PMC13426974", "ISSN": "1932-6203", "publisher": "PLOS", "URL": "https://doi.org/10.1371/journal.pone.0354882", "language": "en", "issued": { "date-parts": [ [ 2026, 7, 31 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1093/bib/bbag411 [code]: Screening of cell-type-specific meta-programs for drug repurposing in Alzheimer's disease.
Journal: Briefings in bioinformatics
In common: genetics / omics, 3 references
[2] doi:10.1371/journal.pone.0352503: G-protein coupled receptor, signal and signal-transduction related transcript mapping in nigral volume and pigmented neurons reveals a potential deficit of nigral feedback signals associated with Parkinson's disease.
Journal: PloS one
In common: genetics / omics, cellular / molecular, 2 references
[3] doi:10.3390/ijms27146196: Investigation of the Potential Neuroprotective Mechanisms of <i>Acalypha indica</i> Against Alzheimer's Disease by Integrated Bioinformatics Analysis.
Journal: International journal of molecular sciences
In common: genetics / omics, cellular / molecular, 2 references
[4] doi:10.1186/s13024-026-00946-0: Insulin resistance alters cortical inhibitory neurons and microglia to exacerbate Alzheimer's knock-in mouse phenotypes.
Journal: Molecular neurodegeneration
In common: genetics / omics, other condition, cellular / molecular, 2 references
[5] doi:10.3390/metabo16080541: Spatial Metabolomics and Single-Cell Virtual Knockout Screening Reveal Solanesol Improves Parkinson's Disease-like Pathology Based on Lipid Inflammation Mechanism.
Journal: Metabolites
In common: genetics / omics, cellular / molecular, 2 references
[6] doi:10.3390/cimb48040369: Elucidating the Multi-Target Anti-Pruritic Mechanism of <i>Polygonatum odoratum</i> via Integrated Network Pharmacology, Molecular Simulations, and GEO Dataset Validation.
Journal: Current issues in molecular biology
In common: cellular / molecular, 2 references
[7] doi:10.1371/journal.pone.0347532: Deciphering the molecular network of Trichostatin A in regulating Alzheimer's disease screening of core genes and mechanistic investigation based on multidimensional bioinformatics and molecular simulation.
Journal: PloS one
In common: 2 references
[8] doi:10.1186/s13041-026-01313-2: REST deficiency and neurogenic-to-gliogenic shift in down syndrome human cerebral organoids.
Journal: Molecular brain
In common: genetics / omics, other condition, 1 reference
[9] doi:10.1186/s40035-026-00567-z: Transaldolase 1 contributes to pentose phosphate pathway disruption and synaptic dysfunction in Alzheimer's disease.
Journal: Translational neurodegeneration
In common: genetics / omics, cellular / molecular, 1 reference
[10] doi:10.1242/dmm.052585: Induced pluripotent stem cells from a transgenic minipig model of Huntington's disease reveal early metabolic changes.
Journal: Disease models & mechanisms
In common: other condition, cellular / molecular, 1 reference

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.