Hadi Azarabad - Bioinformatics | Single-cell & Multi-omics | Reproducible Pipelines | Interpretable ML
I build reproducible computational biology workflows and models across single-cell and multi-omics, with a current focus on immune signaling and cancer biology.
- Multi-omics integration (bulk RNA-seq + ChIP-seq + scRNA-seq) for mechanistic inference (e.g., STAT1 / interferon response)
- Single-cell multiome analysis and gene regulatory network inference
- Clinically aligned ML with interpretability and error analysis (e.g., AML cytomorphology)
- Nextflow multi-omic STAT1 Response Atlas — end-to-end workflow integrating RNA/ChIP/scRNA; mapped 800 STAT1 peaks → 710 genes and identified 42 interferon-responsive genes
https://github.com/mhazarabad/Nextflow-practice - PBMC Multiome Regulatory Network — GRN inference from 10x PBMC multiome; AP-1 (FOS/FOSB) and STAT1 prioritized as key regulators
https://github.com/mhazarabad/PBMC-Multiome-Regulatory-Network - AML Cytomorphology — STN + ResNet-18 model, 94.7% accuracy across 15 cell types; interpretable attention + clinical error analysis
https://github.com/mhazarabad/AML-Cytomorphology - Single-Cell DNA CNV Detection (CopyKAT Benchmark) — HCC1954 CNV calling with automated QC; reproduces chr1q/chr3 gains
https://github.com/mhazarabad/scDNA-Analysis - CRISPR gRNA Design Benchmark (HPRT1) — rule-based design + validation vs Doench 2016 & Azimuth
https://github.com/mhazarabad/Crispr-gRNA-Design
- Reproducibility first (parameterized pipelines, versioned environments, small test runs)
- Transparent evaluation (ablations, error analysis, and documented limitations)
- Code that can be reused by other researchers
- Email: mhazarabad@gmail.com
- LinkedIn: https://www.linkedin.com/in/mhazarabad/
- Location: Pisa, Italy