Rafael Soler Ortuño · Computational Biologist
I am Rafael, a computational biologist who connects single-cell, spatial and multi-omic data with interpretable machine learning to support target discovery, biomarker strategy and patient stratification.
Curious by nature and rigorous by training, I enjoy the difficult space between data, mechanism and therapeutic decisions. I build analyses and tools that make the evidence visible—not just the final answer.
Based in Lund, Sweden · Open to opportunities from September 2026
01 / Now
BioInvent International AB · Lund, Sweden
Contract period Aug 2025 — Aug 2026
Company focus Antibody therapeutics for oncology and autoimmune diseases
Role Multi-omics lead · Immunotherapy · Biomarker discovery · Patient stratification · Mode of action
I lead computational work across antibody programs, integrating clinical, mouse and in-vitro evidence to resolve response and resistance mechanisms. My day-to-day spans scRNA-seq, CITE-seq, spatial transcriptomics, proteomics, clinical metadata and reproducible AWS/HPC workflows.
02 / Selected tools
Eight public projects spanning target discovery and safety, method reliability, pharmacogenomics, safe agentic development and explainable protein-sequence modelling.
| Field | Tool | The problem | What it is for | What it adds | Key skills applied |
|---|---|---|---|---|---|
| Target discovery & safety | TargetIntel-IO | A gene linked to disease is not automatically a good drug target, yet most tools collapse that judgement into a single score. | Sorting candidate genes into genuine targets, biomarkers, resistance mechanisms and poor candidates. | Three separate rankings, one per therapeutic strategy, plus the evidence for and against each call. |
|
| Activation Liability | A target can look safely tumour-specific in healthy resting tissue and stop looking that way once the tissue is inflamed. | Spotting targets whose apparent safety margin shrinks under immune activation. | A direct resting-versus-activated comparison in public data, with abstention when the evidence is too thin. |
|
|
| reach-gap | A target can be highly expressed and still sit too far from a working blood vessel for an antibody to reach it. | Asking whether target-positive tumour cells are physically accessible to an antibody. |
Mechanistic diffusion modelling paired with an evidence graph that returns NOT_COMPUTED rather than guessing.
|
|
|
| Method reliability | segmentation-fragility | In spatial data, a small change in where a cell boundary is drawn can move a marker between cell types and flip the conclusion. | Testing whether a spatial-transcriptomics claim survives a change in transcript-to-cell assignment. | Each claim is labelled robust, segmentation-sensitive or not reportable, instead of trusting one segmentation as final. |
|
| FM Value Audit | A foundation-model embedding can look biologically structured and still add nothing to a real decision. | Testing whether single-cell foundation models beat cheaper baselines at target prioritisation. | A leakage-aware temporal holdout designed to report a negative result when that is what the data shows. |
|
|
| Drug response prediction | DrugMatch-Confidence | Cancer models respond very differently to the same drug, and predictions rarely say how much to trust them. | Predicting how a preclinical cancer model will respond to a given drug. | Every prediction carries a calibrated probability, an uncertainty interval and a warning when the input is unfamiliar. |
|
| Protein deep learning | MTF Prediction | Whether a membrane protein carries transcription-factor-like features can change depending on where you draw its boundary. | Screening human and mouse membrane-protein domains for those features under different boundary definitions. | The same screen repeated across three boundary conditions, so a result that holds under only one is visible as such. |
|
| Research engineering | Dev Autopilot | AI coding agents can move fast through a repository and leave no reliable trail of what changed or why. | Running AI agents on real code with hard limits on what they may touch. | Separate agents implement, audit and review; deterministic gates decide whether work advances, and a person keeps merge authority. |
|
01 v0.6.0 · Research preview
Explainable target discovery and prioritisation
Separates therapeutic targets, biomarkers, resistance mechanisms and poor direct targets in anti-PD-1-resistant melanoma—then ranks them according to antibody/IO, biomarker or small-molecule intent.
3 intent rankings 331 discovery universe 26Q1 real DepMap layer
02 v0.5.0 · Research preview
Hidden safety risk in surface targets
Audits how much of a target’s apparent tumour selectivity disappears when healthy tissue is inflamed rather than resting, using paired public single-cell cohorts.
52 automated tests 85.4% non-I/O coverage
03 v0.8.0 · Research preview
Is the target physically reachable?
Couples simplified reaction–diffusion modelling with an explicit evidence graph to ask whether target-positive tumour cells are accessible to an antibody, and returns NOT_COMPUTED when the required biology has not been measured.
117 automated tests 20,000 uncertainty draws
04 v0.5.0 · Research preview
Does the conclusion survive the segmentation?
Tests whether a spatial-transcriptomics claim still holds when transcript-to-cell assignment changes across Xenium, ProSeg and Baysor, labelling each claim robust, segmentation-sensitive or not reportable.
62 automated tests 90.3% line coverage
05 v0.5.0a1 · Pre-release
Do foundation models actually add value?
A leakage-aware temporal-holdout benchmark testing whether Geneformer and scGPT representations improve target prioritisation over cheaper baselines. The completed v0.4 result was negative, and is reported as such.
79,542 holdout rows NOT_SUPPORTED primary conclusion
06 v1.0.1 · Released
Interpretable cancer drug-response prediction
Predicts sensitivity in preclinical cancer cell lines from DepMap multi-omics and PRISM response data, with SHAP drivers, calibrated uncertainty, prediction intervals and out-of-distribution warnings.
5 real drug models 2 validated demos GDSC2 external test
07 v1.0.0 · Research analysis
Topology-aware protein deep learning
Screens human and mouse membrane-protein domains for transcription-factor-like sequence features using DeepTFactor, then interprets selected predictions with Captum Integrated Gradients.
08 v0.1.1 · Alpha
Auditable AI-agent workflow for scientific software
A persistent Python/SQLite orchestrator that separates implementation, adversarial audit and independent review, with deterministic gates and final human authority.
03 / Experience
Six-plus years working across industry and academia, with a consistent focus on translating complex molecular data into useful biological conclusions.
Computational lead for translational multi-omics across anti-TNFR2 and anti-FcγRIIB antibody programs.
Precision oncology, preclinical-to-clinical model mapping and spatial transcriptomics across tumour cohorts.
Cross-species single-cell and epigenomic analysis of cortical development, plus paediatric brain-tumour genomics.
Hi-C, chromatin accessibility and sequence-to-signal deep learning with ChromBPNet.
04 / What I bring
Immuno-oncology, inflammation, tumour biology, mechanism of action, target validation, biomarkers and patient stratification.
scRNA-seq, CITE-seq, spatial transcriptomics, bulk RNA-seq, epigenomics, proteomics and functional genomics.
XGBoost, survival analysis, variational autoencoders, foundation models and sequence-to-signal deep learning, with uncertainty and provenance.
Python, R, Bash, Nextflow, Snakemake, Docker, AWS and HPC—with versioned inputs, automated tests and stakeholder-ready reporting.
05 / Selected publications
06 / Contact
I am interested in computational biology and translational data-science roles where rigorous analysis can shape therapeutic decisions.