Computational oncology researcher · miRNA biomarkers · pancreatic cancer · kidney disease · deep learning histopathology
Based in Beirut, Lebanon · Faculty of Medicine, American University of Beirut · mkm25@aub.edu.lb
I write reproducible, open-source pipelines for biomarker discovery and functional interpretation in solid tumors, primarily pancreatic cancer and renal cell carcinoma, using multi-omics data, machine learning, and clinical cohort analysis.
| Repository | What it does |
|---|---|
| GPRC5A-paradox-PDAC | 5-aim pipeline resolving the GPRC5A prognostic paradox in pancreatic cancer via subtype stratification, treatment deconfounding, CPTAC proteomics, AlphaFold2, and ML role-state classification · Research Square 2025 |
| RNA-harmonization-AI | Batch-harmonized ML framework for cross-cohort RNA biomarker discovery in PDAC · bioRxiv 2025 |
| mirBottleneck | Bioconductor R package scoring miRNAs by transcriptome-stabilizing activity (VSS + coherence induction). Composite bottleneck index predicts OS in TCGA-PAAD (HR=6.55, p=1.43×10⁻⁸, C-index=0.699) · manuscript in preparation |
| miRNA-diagnostic-platform | R Shiny platform for clinical miRNA qPCR analysis in RCC — ΔΔCt, geNorm stability, Wilcoxon testing, ROC-AUC, elastic net multi-miRNA classifier |
| TCGA-KIRC-vs-KIRP-miRNA-Random-Forest-Classifier | Random Forest classifier for kidney cancer subtype discrimination (KIRC vs KIRP) using TCGA miRNA expression data |
| trichrome-analyzer | AI-powered quantification of interstitial fibrosis from trichrome-stained kidney biopsy images |
| kidney-fibrosis-grader | ResNet-FPN deep learning classifier for kidney biopsy fibrosis grading (4 classes) with LLM-generated pathology reports — Streamlit app |
| cDNA-Synthesis-Calculator | Interactive calculator for cDNA synthesis reaction setup |
Languages: R · Python
Bioinformatics: TCGAbiolinks · DESeq2 · survival · limma · edgeR · sva · Bioconductor
ML / DL: caret · xgboost · randomForest · glmnet · pROC · PyTorch · scikit-learn
Apps: R Shiny · Streamlit
Other: ggplot2 · AlphaFold2 · GDC API · CPTAC Data Portal · Zenodo
I am a Christian first and a researcher second. That is not a disclaimer, it is the most honest explanation of why I do this work and what I think it is capable of doing. The history of medicine, read carefully, is a history of people who were not supposed to succeed, succeeding. Ben Carson. Francis Collins. James Allison. People who went through years of difficulty and came out the other side carrying something that could help millions. I do not think that pattern is accidental. I think God works through people, and I think the preparation, including the hard years, is part of the making of an instrument precise enough to do something that needed doing. The question I am most interested in is the one oncology circles around without quite landing on. Not what is cancer. Not how does it resist treatment. But why? why does it take the people it takes? why does it take the innocent? and what does the answer to that question mean for how we fight it? That is a question that cannot be answered by biology alone. But biology is where I can contribute. I wrote about this at length here: On Faith, Suffering, and the Geometry of Disease
Markarian MB. Resolving the GPRC5A Prognostic Paradox in Pancreatic Ductal Adenocarcinoma. Research Square. 2025. doi:10.21203/rs.3.rs-9237732/v1
Markarian MB. Batch-harmonized machine learning framework for cross-cohort RNA biomarker discovery in pancreatic adenocarcinoma. bioRxiv. 2025. doi:10.1101/2025.11.14.688421
Markarian MB. mirBottleneck: A Dual-Score Framework for Identifying Transcriptome-Stabilizing miRNAs and Predicting Survival in Pancreatic Adenocarcinoma. Manuscript in preparation. 2026.