|
Platform Overview
Table of Contents
1. Introduction
OmicsAnalyst 2.0 is a comprehensive web-based platform designed to support end-to-end analysis of multi-omics data. The platform bridges the gap between exploratory statistics and mechanistic interpretation, enabling researchers to move seamlessly from pattern discovery to functional insight and causal hypothesis testing.
OmicsAnalyst 2.0 operates through three complementary analytical components that are integrated within a unified workflow:
- Comprehensive Statistical Integration - Explore patterns and covariance through individual, pairwise, and multi-omics joint analysis
- Interpretable Network Integration - Visualize omics signatures in biological context across omics layers
- Causal Mechanistic Analysis - Elucidate regulatory mechanisms and driver features
2. System Requirements
OmicsAnalyst 2.0 is a web-based application that runs in your browser. For optimal performance, we recommend:
| Component | Requirement |
|---|---|
| Browser | Chrome, Firefox, Safari, or Edge (latest versions) |
| Internet Connection | Stable broadband connection recommended |
| Screen Resolution | 1280x800 or higher |
| JavaScript | Must be enabled |
3. Supported Omics Types
OmicsAnalyst 2.0 supports the following five omics data types for multi-omics integration analysis. A maximum of 5 omics data types can be uploaded per analysis:
Genes/mRNAs
Gene expression data from RNA-seq or microarray experiments. Supported ID types: Entrez ID, Ensembl Gene ID, Ensembl Transcript ID, Official Gene Symbol.
Proteins
Protein abundance data from mass spectrometry-based proteomics. Supported ID types: UniProt Protein ID, Entrez ID, Ensembl Gene ID, Official Gene Symbol.
miRNAs
MicroRNA expression data. Supported ID types: miRBase mature ID, miRBase accession, miRBase ID (e.g., hsa-miR-21).
Metabolites
Metabolite concentration data from metabolomics studies. Supported ID types: KEGG ID, PubChem ID, HMDB ID, Common Name.
Microbiome
Taxonomic abundance data from 16S rRNA or metagenomics. Supported ID types: Taxonomy label, OTU ID. Taxon levels: Phylum, Class, Order, Family, Genus, Species, Strain.
4. Input Data Formats
OmicsAnalyst 2.0 provides two upload modes for different analysis workflows:
Upload Mode 1: Data Tables (Statistical Integration)
Upload expression/abundance data tables with a metadata file for comprehensive statistical analysis.
- Upload a single metadata file and at least two omics data tables
- The metadata table should describe the same sample IDs shared across all omics data
- A small percentage of missing values are acceptable
Upload Mode 2: Feature Lists (Network Integration)
Upload pre-identified feature lists from external analysis tools ( limma, DESeq2, edgeR, etc.) for network-based analysis.
- Upload one or more feature lists (genes, proteins, metabolites, etc.)
- Specify the model organism (Human, Mouse, or Other)
- For microbiome data, specify its host (currently Human or Mouse only)
Supported File Formats
| Format | Extension | Max Size |
|---|---|---|
| CSV | .csv | 50 MB (Genes/mRNAs), 25 MB (others) |
| TXT | .txt | 50 MB (Genes/mRNAs), 25 MB (others) |
| TSV | .tsv | 50 MB (Genes/mRNAs), 25 MB (others) |
Data Table Structure
Each omics data table should follow this structure:
- First column: Feature identifiers (gene symbols, protein IDs, metabolite names, etc.)
- Subsequent columns: Sample measurements (one column per sample)
- First row: Header with sample names matching the metadata file
Metadata File Structure
A metadata file describing sample information is required for Statistical Integration:
- First column: Sample names (must match column headers in data files)
- Second column: Primary study factor (group/condition) - no missing values allowed
- Additional columns: Other sample attributes (batch, clinical variables, etc.)
- Ensure sample names are consistent across all data files and the metadata file
- Avoid special characters in sample names and feature IDs
- No missing values allowed for the primary study factor (first column after sample names)
5. Key Features
OmicsAnalyst 2.0 bridges the gap between exploratory statistics and mechanistic interpretation through three key analytical components:
Statistical Integration
Comprehensive multi-omics statistical analysis including limma-based differential analysis, correlation analysis, and advanced integration methods (MCIA, MOFA, DIABLO) to identify significant features and patterns across data types.
Network Integration
Project significant features onto comprehensive molecular networks, including protein-protein interaction, metabolic pathways, and regulatory networks for biological context and functional interpretation.
Causal Analysis
Test mechanistic hypotheses through mediation analysis and interaction modeling (IntLIM) to identify regulatory relationships and driver features across omics layers.
Statistical Integration Methods
The Statistical Integration component provides a progressive analytical framework from individual omics characterization to full multi-omics integration:
Single-Omics Characterization - Analyze each omics layer individually:
- Significant Features: Identify features significantly associated with experimental factors using linear models (limma)
- Overall Patterns: Explore sample clustering, grouping patterns, and major sources of variation
- Variance Partitioning: Decompose global and feature-level variance to identify features driven by specific factors
Pairwise Omics Analysis - Discover relationships between two omics layers:
- Clustering Heatmap: Hierarchical clustering to reveal global correlation patterns between two omics layers
- Correlation Network: Network-based visualization of significant feature-to-feature correlations
- Differential Chord Diagram: Compare correlation structures between conditions to identify network rewiring
- Sparse CCA: Maximize correlation between two blocks and select key features driving the link
Multi-Omics Integration - Integrate all omics layers simultaneously:
- Global Exploration: Consensus PCA and Multiple Co-inertia Analysis (MCIA) for visualizing shared trends
- Latent Factor Discovery: Fast NMF, Semi-NMF, and MOFA for discovering biologically interpretable patterns
- Feature Selection: DIABLO for finding discriminative and correlated features across omics
Network Integration Methods
For feature list inputs, build and explore comprehensive molecular networks:
- Database Selection: Choose from protein-protein interaction, metabolic, regulatory, and other network databases
- Network Builder: Construct multi-omics networks connecting genes, proteins, metabolites, and microbiome
- Network Viewer: Interactive visualization with functional enrichment analysis
Causal Analysis Methods
Test mechanistic hypotheses and identify regulatory relationships:
- Pairwise Linear Model (IntLIM): Test for significant linear relationships between paired features across conditions
- Mediation Analysis: Identify mediating relationships where one omics layer influences another through an intermediate
6. Analytical Workflows
OmicsAnalyst 2.0 provides two entry points depending on your input data type:
Data Tables → Statistical Integration
Feature Lists → Network Integration
Example Analytical Pipelines
Based on these workflows, users can design their own analytical strategies to address specific biological questions. Below are two example pipelines demonstrating how the three key components work together:
Discovery Workflow: Statistics → Networks → Function
Validation Workflow: Networks → Statistics → Causality
7. Getting Started
To start using OmicsAnalyst 2.0, visit the Upload page and choose your data input type. Before uploading, ensure your data files are properly formatted as described in Section 4: Input Data Formats.
- Data Tables: Choose this if you have raw expression/abundance data and want to perform comprehensive statistical analysis from scratch
- Feature Lists: Choose this if you have already identified significant features using external tools and want to focus on network-based interpretation
