OP Scientific

Independent scientific consulting in computational systems biology and multi-omics

I help teams in academia and industry turn complex biomedical datasets into meaningful insights. That means performing reproducible bioinformatics analysis with clear logic & interpretation within the scope of their biomedical question.

For demanding projects, I offer services to form and coordinate a computational team from my network of experts (e.g., in omics, imaging datasets, AI, drug discovery).

We will find a good solution tailored to your needs - reach out.

Portrait of Oleksandr Petrenko

Oleksandr Petrenko, MD

Vienna, Austria

Email Oleksandr View services

Services

Scientific advisory, study design and training

When you want the study to be analysis-ready from day one - or you need a clear plan to triage and salvage an existing dataset.

  • Omics-aware study design: endpoints, confounders, batch strategy, and "what will we conclude if X happens?"
  • Practical power/feasibility input (including what not to do with the sample size you have)
  • Methods + analysis sections for grants/proposals and internal documents
  • Reviewer-response support: targeted re-analysis, sensitivity checks, and figure refits
  • Hands-on training/workshops using your own data and your preferred stack

Typical format: short consults, project "audit" days, or recurring advisory (e.g., 1-2 days/month).

Deliverables: a decision-oriented analysis plan, written notes, and (when relevant) training materials.

Multi-omics and spatial data analysis

From raw data to figures and tables you can publish or make decisions from - end-to-end or as a focused module.

  • Bulk RNA-seq; sc/snRNA-seq (QC, annotation, differential states, compositional effects)
  • Spatial transcriptomics (10x Visium / Visium HD: QC, deconvolution/colocalization, region-level biology)
  • LC-MS proteomics / metabolomics (QC, differential abundance, pathway context)
  • Multi-omics integration and network/pathway-level interpretation
  • Reproducible pipelines (Snakemake/containers where they add real portability)

Typical format: a scoped analysis module or full workflow with checkpoints.

Deliverables: reproducible code, clean tables, publication-ready figures, and a concise written results narrative aligned to your question.

Translational biomarker and target discovery

For programs that need mechanism, stratification, or actionable signals - not just lists of differentially expressed features.

  • Phenotype-linked signatures and candidate panels designed for follow-up validation
  • Network/module analysis tied to readouts (disease severity, hemodynamics, outcomes)
  • Predictive modeling with clear validation logic and interpretable outputs
  • Target/biomarker prioritization with a transparent rationale and next-step experiments

Typical format: focused discovery sprint or embedded support within a translational program.

Deliverables: ranked candidates/panels, model outputs with validation summaries, and concrete next-step recommendations.

About me

Combination of clinical and computational systems biology expertise (omics -> mechanisms -> translation)
Comfortable with "non-perfect", real-world and nontrivial datasets
Reproducibility by default: I provide versioned code, documented environments, and audit trails
Direct communication and fast integration into existing teams and infrastructure

How I work

  1. 01

    Intake and alignment

    Scope, success criteria, constraints (timeline, compute, data access), and what decisions the analysis must support.

  2. 02

    Data audit and analysis plan

    Quick QC/feasibility readout, then a written plan: contrasts, covariates, validation strategy, and deliverables.

  3. 03

    Iterative checkpoints

    Short cycles with visible outputs (figures/tables), decision notes, and course-corrections early rather than late.

  4. 04

    Delivery and handoff

    Reproducible repo + walkthrough. You should be able to rerun, extend, and defend the work.

  5. 05

    Support through submission / follow-up

    Targeted additions for rebuttals, extra sensitivity checks, and "one more figure" moments.

Send a short note with dataset type, primary question, and timeline

I'll reply with a proposed scope, what I'd do first, and what expected endpoints are.

Status: currently available for collaborations (last updated: March 2026).
Email Oleksandr Request examples

Selected projects

Transcriptomic signatures of progressive and regressive liver fibrosis and portal hypertension

Focus

Gene programs during fibrosis progression vs early regression, linked to portal hypertension physiology.

Data

Mouse liver RNA-seq across two injury models (CCl4 and TAA) at peak fibrosis and early regression, plus validation in human liver disease transcriptomic datasets.

What I delivered

Progression/regression signatures, phenotype-linked modules/network analysis, and a short list of candidate transcriptomic markers with human-data validation.

Outcome

A coherent progression-to-regression story with candidate markers connected to portal pressure and fibrosis readouts.

Paper

iScience, 2024.

10.1016/j.isci.2024.109301

Metabolomic profiles differentiating porto-sinusoidal vascular disorder and cirrhosis

Focus

Non-invasive differentiation of PSVD vs cirrhosis and pathway signals that may reflect PSVD biology.

Data

Serum LC-MS metabolomics in PSVD, cirrhosis, and healthy controls; differential abundance with limma and classifier models using small metabolite panels, with external validation.

What I delivered

Statistical analysis + feature selection and classification framing, plus interpretable metabolite signatures suitable for follow-up.

Outcome

Compact metabolite signatures (including strong single-marker/ratio performance in validation) that help separate PSVD from controls and (more modestly) from cirrhosis.

Paper

JHEP Reports, 2024.

10.1016/j.jhepr.2024.101208

Non-invasive assessment of portal hypertension severity using machine learning

Focus

Predicting clinically significant portal hypertension and severe portal hypertension from routine laboratory parameters in compensated advanced chronic liver disease.

Data

1,232 participants with cACLD with HVPG measurements; ML models using 3 or 5 routine lab parameters, evaluated across internal and external cohorts; online calculator provided.

What I delivered

Modeling and validation logic around practical lab-only predictors, plus performance context vs established non-invasive testing.

Outcome

Deployable risk stratification for HVPG ≥10 mmHg and ≥16 mmHg using widely available labs, with an accompanying web tool.

Paper

Journal of Hepatology, 2023.

10.1016/j.jhep.2022.09.012

FAQ

What engagement models do you offer?

Scoped projects (including fast "audit + plan" work), ongoing analysis support for active studies, and advisory retainers for decision-heavy programs.

How do you price work?

Either day-rate (useful when the scope is evolving) or fixed-price for a tightly defined module. After a short intake, I'll propose the simplest option.

Do you sign NDAs and handle sensitive data?

Yes. Work can be done inside your environment (your compute + your policies) when required.

What about authorship?

If the contribution meets authorship criteria, I prefer to agree on expectations early. Otherwise, acknowledgement is standard.

What is the typical turnaround time?

A first-pass data audit and proposed plan usually takes days, not weeks. Full timelines depend on data volume, access constraints, and how many iteration cycles you want.

What do you need to start?

Dataset type, sample counts, the primary question, what "success" looks like, and any hard constraints (deadline, compute, access). If you already have a draft design/analysis plan, send it - I'll stress-test it quickly.

Contact

Email oleksandr@opscientific.eu
GitHub github.com/xander-p
ORCID 0000-0002-8586-4910
LinkedIn linkedin.com/in/xander-petrenko
Email Oleksandr Services