Cross-omics association identification is the most critical issue in multi-omics research, yet there is a lack of relevant analytical algorithms and tools. Due to the lengthy measurement processes in omics data generation, experimental errors are quite complex. Moreover, different omics measurements introduce distinct types of errors, making it difficult to apply similar methods for unified correction across multiple omics datasets. These factors severely interfere with cross-omics association analysis. Additionally, current cross-omics association approaches often rely heavily on correlation calculations, failing to understand the relationships between features and overall patterns, as well as important issues in microbiome research such as the biological distribution characteristics of features. This leads to insufficient potential validity and biological verifiability of cross-omics associations.
Our team has long focused on developing methods and tools related to microbiome research. We have previously developed ggClusterNet, EasyAmplicon, EasyMicrobiomeR, and EasyMetagenome tools, which are currently widely recognized and used by peers. These publications have all been classified as highly cited papers. Therefore, after thoroughly analyzing the various difficulties existing in current cross-omics association processes, we developed the OmicsBridge algorithm for mining cross-omics associations. Based on this algorithm, we systematically integrated multi-omics data mining workflows and developed the R package EasyMultiOmics for public use. We hope our tools can accelerate the application of multi-omics in biological scientific research and expedite the discovery of biological associations.
- Provides a comprehensive multi-omics analysis pipeline with emphasis on count data and mass spectrometry data integration.
- Enables bidirectional correlation analysis between count data and mass spectrometry data.
- Incorporates intelligent feature selection with statistical power consideration.
- Offers sophisticated data alignment for batch effect correction and multi-omics integration.
- Generates structured output with publication-ready visualizations and organized result directories.
- /data: This folder contains example datasets used for demonstrating the functionality of the package.
- /R: This folder contains the source code for all the functions in the package. Each file corresponds to a specific analysis or utility function.
- /pipeline: This folder contains the workflow scripts for running different analyses. These scripts provide a structured approach to performing multi-omics data mining.
To install EasyMultiOmics, you can use the following commands in R:
# Install devtools if not already installed
install.packages("devtools")
library(devtools)
# Install EasyMultiOmics from GitHub
devtools::install_github("taowenmicro/EasyMultiOmics")The across_omics.R script is a key component of the R package, designed to perform integrated analysis across different omics datasets. Here’s a detailed breakdown of its functionality:
Start by loading the necessary R packages.
# Load libraries
library(sva) # ComBat batch correction
library(RSpectra) # Sparse PCA
library(dplyr) # Data manipulation
library(compositions) # CLR
library(ggClusterNet)
library(phyloseq)
library(devtools)
load_all()
library(EasyMultiOmics)
library(caret)
library("DESeq2")
library(limma)
library(randomForest)
library(DESeq2)
library(igraph)
library(openxlsx)Create the main directory structure for storing results.
# Create main directory ----
main_dir <- "./result/across_omics"
if (!dir.exists(main_dir)) dir.create(main_dir)Perform feature selection in genomics data using various methods and thresholds.
# Feature selection in genomics (using microbiome data as an example)#------
ps = ps.16s %>% subset_samples.wt("Group",c("WT","KO"))
# ps = ps %>% subset_taxa.wt("Species",c("unclassified"),TRUE)
ps = ps %>% tax_glom_wt(6)
ps
res = count_selection(ps,
group_var = "Group",
prevalence_threshold = 0.1,
detection_threshold = 0.001,
diff_method = "DESeq2",
ml_method = "rf",
cor_method = "spearman",
cor_threshold = 0.6,
p_threshold = 0.05,
weights = list(abundance = 0.3,
importance = 0.3,
differential = 0.2,
network = 0.2))
dat = res$final_ranking %>% head(100)
# Save feature selection results
genomics_wb <- createWorkbook()
addWorksheet(genomics_wb, "genomics_features")
addWorksheet(genomics_wb, "genomics_selection_stats")
writeData(genomics_wb, "genomics_features", dat, rowNames = TRUE)
if(!is.null(res$stats)) {
writeData(genomics_wb, "genomics_selection_stats", res$stats, rowNames = TRUE)
}
saveWorkbook(genomics_wb, file.path(main_dir, "genomics_feature_selection_results.xlsx"), overwrite = TRUE)
Identify metabolites associated with a target microbe.
##Find Associated Metabolites with Target Microbe------
ps2 = ps.ms %>% subset_samples.wt("Group",c("WT","OE"))
tab = ps2 %>% vegan_otu() %>%
as.data.frame()
ps = ps.16s %>% subset_samples.wt("Group",c("WT","OE")) %>% tax_glom_wt(6)
results <- count_to_mass(
phyloseq_obj = ps,
metabolome_mat = tab,
target_microbe_name = "Actinocorallia" ,
top_n = 30
)
names(results)
results$top_metabolites
tax = ps2 %>% tax_table() %>% as.data.frame()
head(tax)
dat = results$top_metabolites %>% left_join(tax,by = c("metabolite"="metab_id"))
# Save microbe-metabolite association results
microbe_metabolite_wb <- createWorkbook()
addWorksheet(microbe_metabolite_wb, "microbe_metabolite_association")
addWorksheet(microbe_metabolite_wb, "association_details")
writeData(microbe_metabolite_wb, "microbe_metabolite_association", dat, rowNames = TRUE)
if(!is.null(results$correlation_matrix)) {
writeData(microbe_metabolite_wb, "association_details", results$correlation_matrix, rowNames = TRUE)
}
saveWorkbook(microbe_metabolite_wb, file.path(main_dir, "microbe_metabolite_association_results.xlsx"), overwrite = TRUE)Perform multi-omics alignment to correct and align different omics datasets.
# Multi-omics alignment
otu = ps.16s %>%
tax_glom_wt(6) %>%
vegan_otu() %>%
as.data.frame()
tab = ps.ms %>% vegan_otu() %>%
as.data.frame()
omics_list = list(micro = otu,ms = tab)
alignment_results <- multi_omics_alignment(
microbiome_data = otu, # Sample x Microbiome
metabolome_data = tab, # Sample x Microbiome
n_factors = 10, # Number of latent factors
method = "fast" # Automatically choose between fast and complete methods
)
# Save alignment results
alignment_wb <- createWorkbook()
addWorksheet(alignment_wb, "alignment_quality")
addWorksheet(alignment_wb, "aligned_microbiome")
addWorksheet(alignment_wb, "aligned_metabolome")
writeData(alignment_wb, "alignment_quality", alignment_results$quality_details, rowNames = TRUE)
writeData(alignment_wb, "aligned_microbiome", alignment_results$aligned_microbiome, rowNames = TRUE)
writeData(alignment_wb, "aligned_metabolome", alignment_results$aligned_metabolome, rowNames = TRUE)
saveWorkbook(alignment_wb, file.path(main_dir, "multi_omics_alignment_results.xlsx"), overwrite = TRUE)Other pipeline scripts in the package follow a similar structure, focusing on specific types of analyses. For example, the 1.1.pipeline.metm.R script performs microbiome analysis.
Start by reading the necessary data and setting up parameters.
rm(list=ls())
# BiocManager::install("MicrobiotaProcess")
library(EasyMultiOmics)
library(phyloseq)
library(tidyverse)
library(ggClusterNet)
library(ggrepel)
library(openxlsx)
# Effective Gene Count
sample_sums(ps.micro)
## Set parameters-----
map= sample_data(ps.micro)
head(map)
phyloseq::tax_table(ps.micro) %>% head()
# Extract the number of unique group factors
gnum = phyloseq::sample_data(ps.micro)$Group %>% unique() %>% length()
gnum
#--Set the order for plotting according to the order in the map file of the ps.micro object
axis_order = phyloseq::sample_data(ps.micro)$Group %>%unique();axis_order
col.g = c("KO" = "#D55E00", "WT" = "#0072B2", "OE" = "#009E73")
#-theme--
package.amp()
res = theme_my(ps.micro)
mytheme1 = res[[1]]
mytheme2 = res[[2]];
colset1 = res[[3]];colset2 = res[[4]];colset3 = res[[5]];colset4 = res[[6]]Perform various analyses, such as alpha and beta diversity calculations.
# alpha diversity -----
#1 alpha.metm:#----
all.alpha = c("Shannon","Inv_Simpson","Pielou_evenness","Simpson_evenness" ,"Richness" ,"Chao1","ACE" )
#--Calculate alpha diversity metrics
tab = alpha.metm(ps = ps.micro,group = "Group" )
head(tab)
data = cbind(data.frame(ID = 1:length(tab$Group),group = tab$Group),tab[all.alpha])
head(data)
data$ID = as.character(data$ID)
result = MuiKwWlx2(data = data,num = 3:6)
result1 = FacetMuiPlotresultBox(data = data,num = 3:6,
result = result,
sig_show ="abc",ncol = 4,width = 0.4 )
p1_1 = result1[[1]] +scale_fill_manual(values = col.g)
p1_1+
ggplot2::scale_x_discrete(limits = axis_order) +
# theme_cell()+
theme_nature()+
ggplot2::guides(fill = guide_legend(title = none))
# Save path
alppath <- file.path(path, "alpha")
dir.create(alppath, showWarnings = FALSE)
# Save Alpha diversity images
ggsave(file.path(alppath, "alpha_diversity_boxplot.png"), plot = p1_1, width = 10, height = 8)
ggsave(file.path(alppath, "alpha_diversity_boxplot.pdf"), plot = p1_1, width = 10, height = 8)