C Proteomic Network-based Characterization of Intrinsic Subtypes of Breast Cancer

C.1 Data description

The PAM50 (Parker et al. 2009) proteomics dataset was obtained from The Cancer Genome Atlas (TCGA, Weinstein et al. (2013)) and The Cancer Proteome Atlas (TCPA,Li et al. (2013)). Reverse Phase Protein Arrays (RPPAs) were used to quantify 190 proteins or phosphoproteins which were involved in apoptosis, DNA damage response, cell cycle (Akbani et al. 2014) over 859 breast cancer (BRCA) patients where normal-like BRCA patients are excluded due to small sample sizes. BRCA of Luminal A and Luminal B are combined in one category to achieve higher power which leads to three groups with 626 BRCA patients in Luminal A and B, 75 patients in Her2-enriched and 158 in Basal-like.

C.2 Preprocessing and application

Proteomics data is standardized before applying GraphR. The indicators for cancer subtype (Luminal A+B, Her2-enriched and Basal-like breast cancers) are used as intrinsic factors.

We set hyperparameters as follows: \(a_\tau = b_\tau = 0.005,a_\pi = 1,b_\pi=4\). Notably, in case of integrated functional pathway analysis, we change the hyper-parameters: \(a_\pi = b_\pi = 0.05\). This provides us potential to consider high density of connections between proteins in the same pathway.

Moreover, we here only include proteins pairs belonging to different isoforms while showing significantly strong correlation (FDR based p-values \(<0.01\) and magnitude \(\mid \rho \mid \geq 0.4\)) for at least one group, allowing plots to be more readable.

C.3 Results

Figure C.1 shows heatmap of posterior inclusion probability (PIP) of selected edges in each PAM50 subtype of BRCA.

Posterior inclusion probability (PIP) of selected edges in each PAM50 subtype of BRCA.

Figure C.1: Posterior inclusion probability (PIP) of selected edges in each PAM50 subtype of BRCA.

Figure C.2, C.3 and C.4 show networks of selected protein pairs in Luminal A+B, Her2-enriched and Basal-like BRCA respectively. The widths of edges are proportional to partial correlations. The sizes of nodes are proportional to connectivity degrees which are defined as the sum of magnitudes of the partial correlations. Sign of partial correlations are represented by color with red being positive and blue being negative.

Luminal A + B breast cancer

Network of selected protein pairs in Luminal A +  B breast cancer patients.

Figure C.2: Network of selected protein pairs in Luminal A + B breast cancer patients.

Her2-enriched breast cancer

Network of selected protein pairs in Her2-enriched breast cancer patients.

Figure C.3: Network of selected protein pairs in Her2-enriched breast cancer patients.

Basal-like breast cancer

Network of selected protein pairs in Basal-like breast cancer patients.

Figure C.4: Network of selected protein pairs in Basal-like breast cancer patients.

References

Akbani, Rehan, Patrick Kwok Shing Ng, Henrica MJ Werner, Maria Shahmoradgoli, Fan Zhang, Zhenlin Ju, Wenbin Liu, et al. 2014. “A Pan-Cancer Proteomic Perspective on the Cancer Genome Atlas.” Nature Communications 5 (1): 1–15.
Li, Jun, Yiling Lu, Rehan Akbani, Zhenlin Ju, Paul L Roebuck, Wenbin Liu, Ji-Yeon Yang, et al. 2013. “TCPA: A Resource for Cancer Functional Proteomics Data.” Nature Methods 10 (11): 1046–47.
Parker, Joel S, Michael Mullins, Maggie CU Cheang, Samuel Leung, David Voduc, Tammi Vickery, Sherri Davies, et al. 2009. “Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes.” Journal of Clinical Oncology 27 (8): 1160.
Weinstein, John N, Eric A Collisson, Gordon B Mills, Kenna R Shaw, Brad A Ozenberger, Kyle Ellrott, Ilya Shmulevich, Chris Sander, and Joshua M Stuart. 2013. “The Cancer Genome Atlas Pan-Cancer Analysis Project.” Nature Genetics 45 (10): 1113–20.