Research
Three lines
of work
Cancer multi-omics, radiogenomics, and the imaging both of them lean on. Beside this panel sits the hub set from my HCC study; hover any gene name to locate it.
Under review
Integrative multi-omics of hepatocellular carcinoma
A single pipeline running from gene retrieval to a survival model. DisGeNET seeds the candidate set, GO and KEGG enrichment pins down the biology, and a STRING network scored in Cytoscape with CytoHubba surfaces the hubs. Four classifiers (SVM, random forest, XGBoost, logistic regression) train on TCGA-LIHC expression, after which LASSO-Cox asks whether the resulting signature carries any prognostic weight.
Under review at a Springer Nature journal · FDR correction applied throughout
In progress
Liver CT radiogenomics
Can the appearance of a lesion on CT tell you anything about what it is doing molecularly? Liver lesions get segmented, radiomic features extracted, and those features tested against molecular subtypes derived from TCGA-LIHC expression. Same cohort as the study above, reached from the imaging side.
Segmentation → radiomic feature extraction → subtype prediction
Published
Peer-reviewed papers