SLCPTAC

Multi-Omics Analysis Toolkit for CPTAC Cancer Database

---

Abstract

SLCPTAC is a comprehensive R package designed for systematic analysis of multi-omics data from the Clinical Proteomic Tumor Analysis Consortium (CPTAC). The package implements 17 analytical scenarios covering correlation analysis, pathway enrichment, and survival analysis across multiple omics layers including transcriptomics, proteomics, phosphoproteomics, genomics, and clinical data.

Key Capabilities

• 17 Analytical Scenarios: Comprehensive coverage of all variable type combinations
• 7 Omics Layers: RNAseq, Protein, Phosphorylation, Mutation, Clinical, Copy Number, Methylation
• 10 Cancer Types: BRCA, LUAD, COAD, CCRCC, GBM, HNSCC, LUSC, OV, PDAC, UCEC
• Automated Workflows: From data loading to publication-ready visualizations
• Phosphoproteomics Focus: Specialized support for phosphorylation site analysis

---

Installation

Prerequisites

# Install Bioconductor packages
if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install(c("fgsea", "ComplexHeatmap"))

Install SLCPTAC

# Install from GitHub
devtools::install_github("SolvingLab/SLCPTAC")

# Install dependencies from SolvingLab
devtools::install_github("Zaoqu-Liu/ggforge")
devtools::install_github("SolvingLab/BioEnricher")
devtools::install_github("Zaoqu-Liu/genekitr2")
devtools::install_github("SolvingLab/astat")
devtools::install_github("SolvingLab/SLCPTAC")

Setup

library(SLCPTAC)

# Set path to CPTAC bulk data
Sys.setenv(SL_BULK_DATA = "/path/to/bulk_data")

---

Analytical Framework

Overview of 17 Scenarios

Scenario	Variable Types	Analysis	Visualization
1	1 continuous vs 1 continuous	Pearson/Spearman correlation	CorPlot, LollipopPlot
2	1 vs multiple continuous	Correlation	LollipopPlot, DotPlot
3	Multiple vs multiple continuous	Correlation matrix	DotPlot
4	1 categorical vs 1 continuous	Wilcoxon/Kruskal-Wallis	BoxPlot
5-6	Multiple BoxPlots	Group comparison	Multiple BoxPlots
7	Categorical vs categorical	Chi-square/Fisher's exact	BarPlot, Heatmap
8-9	1 categorical vs genome/pathways	DEA → GSEA	NetworkPlot, DotPlot
10-11	Multiple categorical vs genome/pathways	Multi-DEA → GSEA	DotPlot Paired, Matrix
12-13	1 continuous vs genome/pathways	Correlation → GSEA	NetworkPlot, DotPlot
14-15	Multiple continuous vs genome/pathways	Multi-correlation → GSEA	DotPlot Paired, Matrix
16	1 variable vs survival	Kaplan-Meier + Cox	KM + Cox curves
17	Multiple variables vs survival	Cox regression	Forest plot

---

Core Functions

1. Data Loading: cptac_load_modality()

Load and merge multi-omics data with automatic preprocessing.

# Load single gene across multiple cancers
data <- cptac_load_modality(
  var1 = "TP53",
  var1_modal = "RNAseq",
  var1_cancers = c("BRCA", "LUAD", "COAD")
)
# Returns: 3 features (TP53 in each cancer)

# Load phosphorylation sites (auto-detected)
data <- cptac_load_modality(
  var1 = "AKT1",
  var1_modal = "Phospho",
  var1_cancers = "BRCA"
)
# Returns: ~9 phospho sites for AKT1

2. Correlation Analysis: cptac_correlation()

Comprehensive correlation and association analysis (Scenarios 1-7).

3. Enrichment Analysis: cptac_enrichment()

Genome-wide scans and pathway enrichment (Scenarios 8-15).

4. Survival Analysis: cptac_survival()

Kaplan-Meier curves and Cox regression (Scenarios 16-17).

---

Methodological Highlights

Correlation Analysis (Scenarios 1-7)

Scenario 1: Transcriptome-Proteome Correlation

Single cancer analysis:

result <- cptac_correlation(
  var1 = "TP53", var1_modal = "RNAseq", var1_cancers = "BRCA",
  var2 = "TP53", var2_modal = "Protein", var2_cancers = "BRCA",
  method = "pearson"
)

Figure 1. Pearson correlation between TP53 mRNA and protein levels in BRCA (r=0.47, p<0.001)

Multi-cancer comparison:

result <- cptac_correlation(
  var1 = "TP53", var1_modal = "RNAseq",
  var1_cancers = c("BRCA", "LUAD", "COAD"),
  var2 = "TP53", var2_modal = "Protein",
  var2_cancers = c("BRCA", "LUAD", "COAD")
)

Figure 2. TP53 mRNA-protein correlation across three cancer types

Scenario 2: Protein-Phosphoproteome Integration

One protein vs multiple phosphorylation sites:

result <- cptac_correlation(
  var1 = "AKT1", var1_modal = "Protein", var1_cancers = "BRCA",
  var2 = c("AKT1", "MTOR", "RPS6"), var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 3. AKT1 protein abundance correlates with downstream phosphorylation events

mRNA vs phosphoproteome:

result <- cptac_correlation(
  var1 = "AKT1", var1_modal = "RNAseq", var1_cancers = "BRCA",
  var2 = "AKT1", var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 4. AKT1 mRNA expression and site-specific phosphorylation patterns

Scenario 3: Phosphorylation Network Analysis

Intra-protein phospho-site correlation:

result <- cptac_correlation(
  var1 = "AKT1", var1_modal = "Phospho", var1_cancers = "BRCA",
  var2 = "AKT1", var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 5. Correlation network among AKT1 phosphorylation sites (diagonal removed)

Cross-protein phosphorylation:

result <- cptac_correlation(
  var1 = "AKT1", var1_modal = "Phospho",
  var1_cancers = c("BRCA", "LUAD", "CCRCC", "UCEC", "PDAC"),
  var2 = "MTOR", var2_modal = "Phospho",
  var2_cancers = c("BRCA", "LUAD", "CCRCC", "UCEC", "PDAC")
)

Figure 6. AKT1-MTOR phosphorylation crosstalk across five cancer types

Proteome-phosphoproteome integration:

result <- cptac_correlation(
  var1 = c("AKT1", "MTOR", "PTEN"), var1_modal = "Protein", var1_cancers = "BRCA",
  var2 = c("AKT1", "MTOR", "RPS6"), var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 7. Systematic protein-phosphorylation correlation matrix in PI3K-AKT-mTOR pathway

Scenario 4: Mutation-Expression Association

Mutation impact on gene expression:

result <- cptac_correlation(
  var1 = "KRAS", var1_modal = "Mutation", var1_cancers = "LUAD",
  var2 = "EGFR", var2_modal = "RNAseq", var2_cancers = "LUAD"
)

Figure 8. KRAS mutation status and EGFR expression levels (Wilcoxon test, p=0.007)

Mutation impact on protein abundance:

result <- cptac_correlation(
  var1 = "TP53", var1_modal = "Mutation", var1_cancers = "BRCA",
  var2 = "AKT1", var2_modal = "Protein", var2_cancers = "BRCA"
)

Figure 9. TP53 mutation status associated with AKT1 protein levels

Multi-cancer mutation-expression analysis:

result <- cptac_correlation(
  var1 = "KRAS", var1_modal = "Mutation",
  var1_cancers = c("LUAD", "COAD"),
  var2 = "EGFR", var2_modal = "RNAseq",
  var2_cancers = c("LUAD", "COAD")
)

Figure 10. KRAS mutation effects on EGFR expression in lung and colon cancers

Scenario 5-6: Mutation Impact on Phosphoproteome

Single mutation vs multiple phospho sites:

result <- cptac_correlation(
  var1 = "PIK3CA", var1_modal = "Mutation", var1_cancers = "BRCA",
  var2 = "AKT1", var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 11. PIK3CA mutation effects on AKT1 phosphorylation sites

Multiple mutations vs phosphoproteome:

result <- cptac_correlation(
  var1 = c("PIK3CA", "TP53"), var1_modal = "Mutation", var1_cancers = "BRCA",
  var2 = c("AKT1", "MTOR", "RPS6"), var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 12. Systematic analysis of mutation effects on PI3K-AKT-mTOR pathway phosphorylation

Multiple phospho sites vs single mutation:

result <- cptac_correlation(
  var1 = c("AKT1", "MTOR", "RPS6"), var1_modal = "Phospho", var1_cancers = "BRCA",
  var2 = "PIK3CA", var2_modal = "Mutation", var2_cancers = "BRCA"
)

Figure 13. Phosphorylation landscape in PIK3CA mutant vs wild-type tumors

Scenario 6: Clinical-Molecular Integration

Clinical variables vs phosphorylation:

result <- cptac_correlation(
  var1 = c("Age", "Tumor_Stage"), var1_modal = "Clinical", var1_cancers = "BRCA",
  var2 = c("AKT1", "MTOR"), var2_modal = "Phospho", var2_cancers = "BRCA"
)

Figure 14. Clinical variables associated with phosphorylation patterns

Tumor stage vs gene expression:

result <- cptac_correlation(
  var1 = "Tumor_Stage", var1_modal = "Clinical", var1_cancers = "BRCA",
  var2 = "TP53", var2_modal = "RNAseq", var2_cancers = "BRCA"
)

Figure 15. TP53 expression levels across tumor stages (Kruskal-Wallis test)

Scenario 7: Co-Mutation and Mutual Exclusivity

Single mutation pair:

result <- cptac_correlation(
  var1 = "KRAS", var1_modal = "Mutation", var1_cancers = "LUAD",
  var2 = "EGFR", var2_modal = "Mutation", var2_cancers = "LUAD"
)

Figure 16. KRAS-EGFR mutual exclusivity in lung adenocarcinoma (percentage stacked bar)

Mutation interaction network:

result <- cptac_correlation(
  var1 = c("KRAS", "EGFR", "ALK", "BRAF"), var1_modal = "Mutation", var1_cancers = "LUAD",
  var2 = c("TP53", "STK11", "KEAP1"), var2_modal = "Mutation", var2_cancers = "LUAD"
)

Figure 17. Mutation co-occurrence and mutual exclusivity landscape. Heatmap shows log2(Odds Ratio): red indicates co-occurrence, blue indicates mutual exclusivity

Clinical-mutation association:

result <- cptac_correlation(
  var1 = "Tumor_Stage", var1_modal = "Clinical", var1_cancers = "BRCA",
  var2 = "PIK3CA", var2_modal = "Mutation", var2_cancers = "BRCA"
)

Figure 18. PIK3CA mutation frequency across tumor stages

---

Enrichment Analysis (Scenarios 8-15)

Scenario 8: Mutation-Driven Proteome Alterations

Genome-wide protein changes:

result <- cptac_enrichment(
  var1 = "KRAS",
  var1_modal = "Mutation",
  var1_cancers = "LUAD",
  analysis_type = "genome",
  genome_modal = "Protein",
  top_n = 30
)

Figure 19. Network visualization of top 50 up/down-regulated proteins in KRAS-mutant tumors

Mutation impact on phosphoproteome:

result <- cptac_enrichment(
  var1 = "TP53",
  var1_modal = "Mutation",
  var1_cancers = "BRCA",
  analysis_type = "genome",
  genome_modal = "Phospho",
  top_n = 30
)

Figure 20. TP53 mutation-associated phosphorylation changes

Scenario 9: Pathway Enrichment Analysis

MsigDB Hallmark gene sets (default, 50 curated pathways):

result <- cptac_enrichment(
  var1 = "PIK3CA",
  var1_modal = "Mutation",
  var1_cancers = "BRCA",
  analysis_type = "enrichment",
  top_n = 20
)

Figure 21. Pathway enrichment in PIK3CA-mutant breast cancer (MsigDB Hallmark)

GO Biological Process enrichment:

result <- cptac_enrichment(
  var1 = "KRAS",
  var1_modal = "Mutation",
  var1_cancers = "LUAD",
  analysis_type = "enrichment",
  enrich_database = "GO",
  enrich_ont = "BP",
  top_n = 20
)

Figure 22. Gene Ontology enrichment for KRAS-mutant lung adenocarcinoma

Reactome pathway analysis:

result <- cptac_enrichment(
  var1 = "TP53",
  var1_modal = "Mutation",
  var1_cancers = "BRCA",
  analysis_type = "enrichment",
  enrich_database = "Reactome",
  top_n = 20
)

Figure 23. Reactome pathway enrichment in TP53-mutant breast cancer

Scenario 10-11: Multi-Variable Enrichment

Multiple mutations genome scan:

result <- cptac_enrichment(
  var1 = c("PIK3CA", "TP53"),
  var1_modal = "Mutation",
  var1_cancers = "BRCA",
  analysis_type = "genome",
  genome_modal = "Phospho",
  top_n = 50
)

Figure 24. Comparative phosphoproteome analysis of PIK3CA and TP53 mutations. DotPlot shows top affected phospho sites for each mutation (red=up, blue=down). Same site may show opposite directions in different mutations.

GSEA matrix for multiple variables:

result <- cptac_enrichment(
  var1 = c("PIK3CA", "TP53"),
  var1_modal = "Mutation",
  var1_cancers = "BRCA",
  analysis_type = "enrichment",
  enrich_database = "MsigDB",
  top_n = 15
)

Figure 25. Pathway enrichment matrix comparing PIK3CA and TP53 mutations

Scenario 12-13: Protein-Centric Enrichment

Protein abundance and phosphoproteome:

result <- cptac_enrichment(
  var1 = "AKT1",
  var1_modal = "Protein",
  var1_cancers = "BRCA",
  analysis_type = "genome",
  genome_modal = "Phospho",
  method = "pearson",
  top_n = 30
)

Figure 26. AKT1 protein abundance correlates with phosphorylation network

Protein expression and pathway activity:

result <- cptac_enrichment(
  var1 = "AKT1",
  var1_modal = "Protein",
  var1_cancers = "LUAD",
  analysis_type = "enrichment",
  enrich_database = "KEGG",
  method = "spearman",
  top_n = 15
)

Figure 27. KEGG pathway enrichment for AKT1 protein expression

Scenario 14-15: Multi-Protein Pathway Analysis

Multiple proteins genome scan:

result <- cptac_enrichment(
  var1 = c("AKT1", "MTOR", "PTEN"),
  var1_modal = "Protein",
  var1_cancers = "BRCA",
  analysis_type = "enrichment",
  enrich_database = "MsigDB",
  method = "pearson",
  top_n = 15
)

Figure 28. Comparative pathway enrichment for PI3K-AKT-mTOR pathway components

---

Survival Analysis (Scenarios 16-17)

Scenario 16: Single Variable Survival

Gene expression and overall survival:

result <- cptac_survival(
  var1 = "TP53",
  var1_modal = "RNAseq",
  var1_cancers = "BRCA",
  surv_type = "OS",
  cutoff_type = "optimal"
)

Figure 29. TP53 expression and overall survival in breast cancer. Left: Kaplan-Meier curve with log-rank test. Right: Cox regression curve showing hazard ratio.

Mutation and survival:

result <- cptac_survival(
  var1 = "KRAS",
  var1_modal = "Mutation",
  var1_cancers = "LUAD",
  surv_type = "OS"
)

Figure 30. KRAS mutation status and survival outcome in lung adenocarcinoma

Phosphorylation and survival:

result <- cptac_survival(
  var1 = "AKT1",
  var1_modal = "Phospho",
  var1_cancers = "BRCA",
  surv_type = "OS",
  cutoff_type = "optimal"
)

Figure 31. AKT1 phosphorylation (mean of all sites) and survival

Scenario 17: Multi-Variable Survival (Forest Plot)

Multiple genes:

result <- cptac_survival(
  var1 = c("TP53", "EGFR", "KRAS"),
  var1_modal = "RNAseq",
  var1_cancers = "LUAD",
  surv_type = "OS",
  cutoff_type = "optimal"
)

Figure 32. Forest plot showing hazard ratios for multiple genes

Multi-cancer comparison:

result <- cptac_survival(
  var1 = "TP53",
  var1_modal = "RNAseq",
  var1_cancers = c("BRCA", "LUAD"),
  surv_type = "OS",
  cutoff_type = "optimal"
)

Figure 33. TP53 expression and survival across breast and lung cancers

Phosphorylation sites survival:

result <- cptac_survival(
  var1 = "AKT1",
  var1_modal = "Phospho",
  var1_cancers = c("BRCA", "LUAD"),
  surv_type = "OS",
  cutoff_type = "optimal"
)

Figure 34. Site-specific phosphorylation and survival across cancers. Each phospho site analyzed independently.

Clinical variables and survival:

result <- cptac_survival(
  var1 = c("Age", "Tumor_Stage"),
  var1_modal = "Clinical",
  var1_cancers = c("BRCA", "LUAD"),
  surv_type = "OS"
)

Figure 35. Clinical variables as prognostic factors across cancers

---

Advanced Applications

Phosphoproteomics-Centered Analysis

SLCPTAC provides specialized support for phosphorylation analysis:

1. Protein-Phospho Correlation:

cptac_correlation(
  var1 = "AKT1", var1_modal = "Protein",
  var2 = "AKT1", var2_modal = "Phospho"
)

2. Cross-Protein Phosphorylation:

cptac_correlation(
  var1 = "AKT1", var1_modal = "Phospho",
  var2 = "MTOR", var2_modal = "Phospho"
)

3. Mutation-Phospho Association:

cptac_correlation(
  var1 = "PIK3CA", var1_modal = "Mutation",
  var2 = "AKT1", var2_modal = "Phospho"
)

4. Phospho Survival Impact:

cptac_survival(
  var1 = "AKT1", var1_modal = "Phospho",
  surv_type = "OS"
)

Multi-Cancer Comparative Analysis

Compare same molecular feature across cancer types:

# Expression correlation
cptac_correlation(
  var1 = "TP53", var1_modal = "RNAseq",
  var1_cancers = c("BRCA", "LUAD", "COAD", "PDAC", "UCEC"),
  var2 = "TP53", var2_modal = "Protein",
  var2_cancers = c("BRCA", "LUAD", "COAD", "PDAC", "UCEC")
)

# Survival comparison
cptac_survival(
  var1 = "TP53", var1_modal = "RNAseq",
  var1_cancers = c("BRCA", "LUAD", "COAD"),
  surv_type = "OS"
)

Pathway-Level Analysis

Multiple enrichment databases supported:

# MsigDB Hallmark (50 gene sets, recommended for quick insights)
cptac_enrichment(..., enrich_database = "MsigDB", msigdb_category = "H")

# GO Biological Process (comprehensive)
cptac_enrichment(..., enrich_database = "GO", enrich_ont = "BP")

# KEGG Pathways
cptac_enrichment(..., enrich_database = "KEGG", kegg_category = "pathway")

# Reactome Pathways
cptac_enrichment(..., enrich_database = "Reactome")

# WikiPathways
cptac_enrichment(..., enrich_database = "Wiki")

---

Data Structure

Feature Labels

Format: "GENE (Modal, Cancer)" or "SITE_GENE (Modal, Cancer)"

Examples:
  - "TP53 (RNAseq, BRCA)"
  - "AKT1 (Protein, LUAD)"
  - "S124_AKT1 (Phospho, BRCA)"
  - "KRAS (Mutation, COAD)"
  - "Tumor_Stage (Clinical, LUAD)"

Column Names

Format: "CancerType_Gene_Modal"

Examples:
  - "BRCA_TP53_RNAseq"
  - "LUAD_AKT1_Protein"
  - "BRCA_S124_AKT1_Phospho"
  - "COAD_KRAS_Mutation"
  - "LUAD_Tumor_Stage_Clinical"

---

Statistical Methods

Correlation Analysis

• Continuous vs Continuous: Pearson/Spearman correlation
• Categorical vs Categorical: Chi-square test or Fisher's exact test, Odds Ratio calculation
• Categorical vs Continuous: Wilcoxon rank-sum (2 groups) or Kruskal-Wallis (3+ groups)

Enrichment Analysis

• Categorical Variables: Differential expression analysis (DEA) using limma
• Continuous Variables: Correlation-based ranking
• GSEA: fgsea with multilevel algorithm
• Multiple Testing: Benjamini-Hochberg FDR correction

Survival Analysis

• Kaplan-Meier: Log-rank test for group comparison
• Cox Regression: Proportional hazards model
• Optimal Cutoff: Maximizes log-rank statistic
• C-index: Concordance index for model performance

---

Supported Data

Omics Layers

Layer	Description	Type	Cancer Coverage
RNAseq	Gene expression (mRNA)	Continuous	All 10 types
Protein	Protein abundance	Continuous	All 10 types
Phospho	Phosphorylation sites	Continuous	8 types*
Mutation	Somatic mutations	Categorical	All 10 types
Clinical	Clinical variables	Mixed	All 10 types
logCNA	Copy number alterations	Continuous	All 10 types
Methylation	DNA methylation	Continuous	7 types**

*Phospho available: BRCA, CCRCC, GBM, HNSCC, LUAD, LUSC, PDAC, UCEC
**Methylation available: CCRCC, GBM, HNSCC, LUAD, LUSC, PDAC, UCEC

Cancer Types

• BRCA: Breast invasive carcinoma
• CCRCC: Clear cell renal cell carcinoma
• COAD: Colon adenocarcinoma
• GBM: Glioblastoma multiforme
• HNSCC: Head and neck squamous cell carcinoma
• LUAD: Lung adenocarcinoma
• LUSC: Lung squamous cell carcinoma
• OV: Ovarian serous cystadenocarcinoma
• PDAC: Pancreatic ductal adenocarcinoma
• UCEC: Uterine corpus endometrial carcinoma

---

Best Practices

1. Phosphorylation Analysis

• Input gene name only (e.g., "AKT1"), sites are auto-detected
• Each phospho site is treated as independent variable
• Use correlation to identify site-specific regulation
• Use survival to find prognostic phospho sites

2. Multi-Cancer Studies

• Use same gene across cancers for direct comparison
• Forest plots automatically generated for multi-cancer survival
• Lollipop/DotPlot for multi-cancer correlation

3. Enrichment Analysis

• Start with MsigDB Hallmark (50 gene sets) for quick insights
• Use GO BP for comprehensive biological process analysis
• Use KEGG for pathway-level interpretation
• top_n controls plot density, stats returns all results
• Clinical variables NOT supported (use correlation instead)

4. Clinical Variables

• Cannot be used in enrichment analysis (multi-category issue)
• Use cptac_correlation() for clinical-molecular associations
• Tumor_Stage, Age are most commonly used
• Gender may have single category in some cancers (e.g., BRCA)

5. Performance

• Large analyses (36 phospho sites × 12000 genes) may take time
• Progress messages provided
• Results cached in slcptac_output/ directory
• Large plots automatically handled (up to 100×100 inches)

---

Citation

If you use SLCPTAC in your research, please cite:

Liu, Z. (2025). SLCPTAC: Multi-Omics Analysis Toolkit for CPTAC Cancer Database.
R package version 1.2.0. https://github.com/SolvingLab/SLCPTAC

CPTAC Network. (2020). Clinical Proteomic Tumor Analysis Consortium.
https://proteomics.cancer.gov/programs/cptac

---

Contact and Support

• Author: Zaoqu Liu, MD, PhD; Yuyao Liu
• Email: liuzaoqu@163.com
• ORCID: 0000-0002-0452-742X
• GitHub: https://github.com/SolvingLab/SLCPTAC
• Issues: https://github.com/SolvingLab/SLCPTAC/issues

License

GPL (>= 3)

Acknowledgments

This work is built upon data generated by the Clinical Proteomic Tumor Analysis Consortium (CPTAC) and The Cancer Genome Atlas (TCGA). We thank all patients, clinicians, and researchers who contributed to these invaluable resources.

---

For detailed tutorials and examples, see: tutorials/Comprehensive_Tutorial.R

Last updated: December 2025