Saber Hafezqorani

github.com/saberhq linkedin.com/in/saberhq x.com/saberhq scholar.google.com

Computational biologist at the intersection of AI × Biology, specializing in large-scale single-cell CRISPR screens. Co-led end-to-end analysis of two genome-wide Perturb-seq screens at Genentech (gRED) and built the team’s benchmarking and dashboard tooling (mapdash); currently competing in Arc Institute’s Virtual Cell Challenge 2026 (Sidechain). Shipped widely adopted methods for NGS data (ntEmbd, NanoSim), with publications in Nature Methods and Nature Communications.

01 Experience
Sep 2025 — Aug 2026
1 year
South San Francisco,
CA, USA (remote)
Computational Scientist Contract Genentech (gRED)
  • —Co-led end-to-end analysis of two multi-million-cell scale genome-wide single-cell CRISPR Perturb-seq screens in breast (MCF7) and lung (A549) cancer cell lines, from data ingest and QC through guide assignment, perturbation-effect estimation (differential expression), embedding, and evaluation against protein complex ground truth
  • —Diagnosed a barcode demultiplexing issue and collaborated with upstream teams to resolve it, recovering approximately 15% more singlet cells for downstream analysis and avoiding costly resequencing
  • —Built mapdash, a CLI tool generating interactive dashboards to evaluate perturbation maps across parameter sweep runs, plus a reusable sweep orchestrator – adopted by the team for systematic benchmarking of DE methods, embedding approaches, and bias correction across Nextflow pipelines on HPC (Slurm)
  • —Benchmarked design choices across the Perturb-seq analysis pipeline – artifact/confounder correction, QC filtering, statistical modeling, and dimensionality reduction – with findings adopted as best practices within internal perturbation map building initiatives
Sep 2017 — Aug 2024
6 years, 11 months
Vancouver, BC, Canada
Doctoral Research Assistant BC Cancer Genome Sciences Centre
  • —Designed and implemented ntEmbd, a deep learning foundation model for nucleotide sequence representation using an unsupervised autoencoder architecture, outperforming state-of-the-art tools in RNA classification tasks
  • —Co-developed NanoSim, a Nanopore multi-omics long-read synthetic sequence generation pipeline (62,000 downloads, 300+ GitHub stars), enabling tool benchmarking across diverse research areas
  • —Co-authored 4 research papers on single-cell transcriptome visualization, long-read sequencing assembly, simulation, and benchmarking, published in Nature Communications and Nature Methods (300+ citations)
  • —Maintained and documented pipelines using workflow management tools, deployed across distributed, HPC (Slurm) and cloud environments (GCP) with robust Git-based versioning, automated testing, and code reviews
  • —Mentored graduate students whose work was presented at international conferences and earned travel awards
Feb 2013 — Aug 2017
4 years, 7 months
Ankara, Turkey
Research Associate & Master’s Research Assistant METU Informatics Institute
  • —Designed a machine learning framework for patient stratification and clinical outcome prediction in neuroblastoma using multimodal data integration (RNA-seq, microarray and aCGH)
  • —Modelled RBP–microRNA interplay in post-transcriptional regulation using CLIP-seq assays, knockdown perturbation experiments, and RNA structure analysis (RNAplfold/Sfold), enabling causal inference of regulatory effects on mRNA stability and gene expression
02 Independent Research
Aug 2026 — Present
1 month
Virtual Cell Challenge 2026 Sidechain
  • —Solo entry predicting transcriptional response to genetic perturbation in unseen cell contexts; evaluation-first design mirroring the official challenge scorer before any modeling
  • —Testing sequence-encoded post-transcriptional priors (RBP–miRNA regulation, 3′UTR grammar) for cross-context generalization — extending my NAR 2016 work into perturbation-response prediction
SABER HAFEZQORANI — RÉSUMÉ01 / 02
03 Education
2017 – 2024
Doctor of Philosophy, Bioinformatics — University of British Columbia (UBC)
Thesis: Machine learning in transcriptome analysis using long RNA sequencing data
2013 – 2015
Master of Science, Bioinformatics — Middle East Technical University (METU)
Thesis: Modeling the combined effect of RBPs and microRNAs in post-transcriptional regulation
04 Selected Publications full list → Google Scholar
BioRxiv
2024
ntEmbd: Deep learning embedding for nucleotide sequences
Hafezqorani et al.  [Paper] [Code] [Video] [Blog]
GigaScience
2020
Trans-NanoSim characterizes and simulates nanopore RNA-seq data
Hafezqorani et al.  [Paper] [Code] [Blog]
Nature Comms
2023
Reference-free assembly of long-read transcriptome sequencing data with RNA-Bloom2
Nip, Hafezqorani et al.  [Paper] [Code]
Nature Methods
2024
Systematic assessment of long-read RNA-seq methods for transcript identification & quantification
LRGASP Consortium, et al.  [Paper] [Code] [Website]
Nucleic Acids Res.
2016
Modeling the combined effect of RBPs and microRNAs in post-transcriptional regulation
Hafezqorani et al.  [Paper]
05 Selected Talks & Presentations
2025 · 2023 · 2019
2018 · 2017 · 2015
scverse (Palo Alto, CA) · ISMB/ECCB — MLCSB, Machine Learning in Computational and Systems Biology (Lyon, France) · ISMB/ECCB — HiTSeq, High Throughput Sequencing Algorithms & Applications (Basel, Switzerland) · Genome Informatics (Hinxton, Cambridge, UK)
+ 10 other academic conference presentations, with attendances at NeurIPS and ICML.
06 Technical Skills
Bioinformatics
& Omics
NGS (long/short RNA-seq, single-cell and bulk transcriptomics, metagenomics), Perturb-seq, CROP-seq, guide assignment & guide-level QC, perturbation-effect estimation, batch/confounder correction, 10X Genomics (3’/5’/FLEX), multi-omics integration & visualization, protein-binding assays (ChIP/CLIP-Seq), RNA-protein interactions, RBPs, microRNAs, cancer genomics, RNA secondary structure
ML & Deep
Learning
Representation learning, transformers, LSTMs, autoencoders, state-space models, Bayesian optimization, hyperparameter tuning (Optuna), feature selection, matrix factorization & gene module analysis, GNN, VAE, PyTorch, TensorFlow, scikit-learn
Statistical
Modeling
Mixture models, kernel density, EM, Markov models, Bayesian inference
AI & Agentic
Workflows
Claude Code, MCP servers, custom skills, context engineering, LangChain, LangSmith
Workflow &
Infrastructure
Nextflow, Slurm, Google Cloud Platform (GCP), Docker, Conda, GitLab/GitHub, CI/CD
Programming &
Scientific SW
Python, R, Bash & Shell scripting, Linux, Matplotlib, Pandas, NumPy, SciPy, scanpy/AnnData (scverse), pertpy
07 Selected Awards & Academic Service
2017 — 2024
Received over $250K in Scholarships, Fellowships, Travel Awards, and Research Fundings
2015 — 2025
Reviewer for Bioinformatics, NAR Genomics and Bioinformatics, Bioinformatics Advances · Active member, International Society for Computational Biology (ISCB)
SABER HAFEZQORANI — RÉSUMÉ02 / 02