Integrative analysis of methylation and transcriptional profiles to - PowerPoint PPT Presentation

Integrative analysis of methylation and transcriptional profiles to predict aging and construct aging-specific cross- tissue network Yin Wang Fudan University

Outline • Introduction • TCGA data • Integrating and Stepwise Age-Prediction pipeline • PPI network • Functional / enrichment analysis • Aging cross-tissue network • Aging pathway interaction network • Conclusion

Introduction • Aging is emerging as an interesting topic, as aging has been shown to be involved in many disorders, such as Parkinson disease, diabetes and cancers. • Profiling patterns of crucial DNA methylation / mRNA markers change with the chronological age • Many predictors have been applied to identify aging biomarkers and analyze aging functions

Our previous study • mRMR method and kNN classifier

Aging cross-talk networks

Introduction Integrative aging networks from multi-scale data (i.e. methylation and expression) have not been constructed entirely in Homo species • Reconstructing molecular networks also gives systematic approaches to deal with multi-scale data in aging analysis

TCGA data and pre-processing • input: methylation and expression; output: clinical data • training data ： BLCA ， BRCA ， HNSC ， KIRC ， LUAD ， THCA ； 216 samples • test data ： KIRP ， LIHC ， PRAD ； 99 samples • svd method was used to assess the sources of inter-sample variation • z-score method

Integrating and Stepwise Age- Prediction method • Lasso regression for methylation data with age • Determine λ value by cross-validation • Regress residuals by gene expression profiles • sort abs(corr) • Determine genes number (PLS) • Determine PLS vectors • 6-fold and 5-fold

Prediction results

Prediction results

Prediction results

PPI network • Data from String database • confidence score >700 （ 0.7 ） • Dijkstra algorithm to find shortest path between • Controlled by permutation test gene p-value ness TP53 1023 0.016* HSP90A 665 0.009* A1 SRC 363 0.086 STA T3 263 0* BMP2 254 0* AKT1 243 0.759 CD8A 235 0* EP300 229 0* HSPA4 221 0* IL6 207 0.018*

functional / enrichment analysis • methylation of GPR45 • expression of CORO6 in kidney • positive regulation of immune system process (GO:0007059, p-value=1.6643e-08, and FDR =1.3308e-05) and cell adhesion molecules (CAMs, p-value=1.4205e-06, and FDR =2.4517e-04) • negative regulation of phosphate metabolic process (GO:0045936, p-value=9.7695e-05, and FDR =0.0157) in kidney renal papillary cell and Antigen processing and presentation pathway (p- value=3.3052e-06, and FDR =0.0006) in thyroid

Cross-tissue network • young age: ≤50; old age: ≥60 (more than 3 samples), 7 tissues and 21 tissut-tissue pairs • profiles were discretized using two thresholds mean+/-std • Kolmogorov-Smirnov (K-S) value of cumulative distribution between different tissues • absolute difference of K-S value between old and young group was set as the edge (>0.95)

Cross-tissue network • 31 pairs in 6 tissue- tissue • GATA4→EGFL7 shares 4 GO terms (development process) • positive regulation of caspase activity (GO:0043280, p- value=9.8594e-05, and FDR =0.0717)

pathway interaction network • KEGG pathway p-value<0.05 and FDR<0.25 • 7 tissues • summarizing K-S values between pathways from different tissues

pathway interaction network • sum of absolute K-S value differences (>0.6)

sum of all absolute K-S value differences

FDR<0.1

Conclusion • Cellular senescence control aging in immunosenescence theories • Cell adhesion cascades, cell cycle and neurotrophin pathway played important roles in the aging process altogether • head / neck and kidney

Summary • The pipeline find key aging markers with high accuracy • Network analysis (PPI, cross-tissue and pathway interaction) revealed coordinated aging patterns

Acknowledgement • Prof. Yixue Li, Lei Liu, and Lu Xie • Associate Prof. Tao Huang • Thank you !