Skip to content
View Ekin-Kahraman's full-sized avatar
6 followers · 51 following
  • UEA · Huang Lab (Mount Sinai, remote)
  • Norwich, UK

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLO

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: YOLO

Highlights

  • Pro

Block or report Ekin-Kahraman

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Ekin-Kahraman/README.md

Ekin Kahraman

Building rustscenic, a Rust + PyO3 rewrite of SCENIC+, with the Kuan-lin Huang Lab at Icahn Mount Sinai.

Current focus

rustscenic (v0.4.0 release): full SCENIC+ pipeline for single-cell regulatory network analysis. GRN, AUCell, topics, cisTarget, enhancer links, eRegulons.

Benchmarks vs the reference Python stack on identical input (v0.3.x measurements; v0.4.x cross-dataset sweep underway):

  • AUCell 88× faster than pyscenic at per-cell Pearson 0.99 (0.21 s vs 18.6 s on 10x Multiome, 10k cells × 1,457 regulons)
  • GRN 1.78× faster than pinned arboreto 0.1.6 on PBMC (per-TF Spearman 0.63)
  • ~6.3× less memory at 100k cells × 20k genes × 4 stages (6.3 GB vs >40 GB reported for scenicplus)
  • 5 runtime dependencies (numpy, pandas, pyarrow, scipy, anndata) vs 40+ for the reference stack
  • 9/9 cortex TFs recovered on mouse brain E18 multiome (biological validation)

Selected work

  • 7 merged scverse PRs across scanpy and PyDESeq2; open AnnData concat API PR
  • 1,773 DE genes (Zenodo DOI): DESeq2 on SARS-CoV-2 nasopharyngeal RNA-seq, n = 60 primary, 99.8% concordant with n = 484 sensitivity (repo)
  • r = 0.954 on 5-fold CV: PyTorch deconvolution of 484 bulk RNA-seq samples into 14 airway cell types, upper bound with no batch effects (repo)
  • Live deploy: SafetyNett, AI safety-netting prototype for NHS GP workflows, OpenClaw Clinical Hackathon (repo)

evk23umu@uea.ac.uk

Pinned Loading

  1. rustscenic rustscenic Public

    Rust + PyO3 reimplementation of the SCENIC+ single-cell regulatory-network pipeline. ATAC preproc, peak calling, GRN, AUCell, topics, cistarget, eRegulon — one pip install, no Java, no MACS2, no dask.

    Python 7

  2. bulk-rnaseq-differential-expression bulk-rnaseq-differential-expression Public

    Reproducible bulk RNA-seq pipeline for SARS-CoV-2 host response in R (DESeq2, pathway enrichment, viral load and sex-interaction analyses). Zenodo DOI.

    R 1

  3. covid-airway-deconvolution covid-airway-deconvolution Public

    PyTorch deconvolution of 484 COVID-19 nasopharyngeal samples into 14 airway cell types using a tissue-matched scRNA-seq reference. Validation r = 0.954.

    Python

  4. single-cell-rnaseq-immune-profiling single-cell-rnaseq-immune-profiling Public

    End-to-end single-cell RNA-seq immune cell profiling pipeline in Python (scanpy, PBMC 3k)

    Python

  5. rnaseq-nextflow-pipeline rnaseq-nextflow-pipeline Public

    Bulk RNA-seq Nextflow pipeline: FastQC, fastp, HISAT2, featureCounts, DESeq2, MultiQC. Dockerised, tested, reproducible.

    Nextflow

  6. safetynett safetynett Public

    AI-powered clinical safety netting for NHS primary care. Red flag detection, automated patient follow-up, GP escalation.

    TypeScript 1 1