BioInvent International AB
Bioinformatics Research Scientist
Single-cell & spatial omicsImmuno-oncologyAutoimmune diseaseAntibody drug programmesBiomarker discoveryTarget validationAWS & HPCAgentic AI
- Ran the multi-omic analysis for the BI-1206 anti-FcγRIIB and BI-1808 anti-TNFR2 antibody programmes. The evidence covers scRNA-seq, CITE-seq, spatial transcriptomics, bulk RNA-seq, proteomics and clinical metadata, tied together across patient, mouse and in-vitro samples, and the questions are mode of action and resistance
- Traced an unexpected immune activation mechanism behind an antibody response from clinical single-cell data, and extended the mode-of-action pipeline to handle that kind of patient cohort
- Characterised drug mechanism of action with scRNA-seq, spatial transcriptomics and targeted gene-module scoring, which resolves antibody-responsive immune cell states, macrophage polarisation profiles and tissue niches in blood and tumour microenvironments. All of that feeds into early patient selection
- Validated cell-type-specific genetic regulation of a candidate autoimmune target using public single-cell data, auditing the raw data quality first so the numbers could be trusted. The result fed a go/no-go decision on the target
- Built biomarker and target evidence for autoimmune and inflammatory disease by curating and integrating public single-cell and spatial atlases across a broad range of conditions: rheumatologic disease, including rheumatoid arthritis with a focus on synovial pathology, systemic lupus erythematosus, juvenile idiopathic arthritis and juvenile dermatomyositis; gastrointestinal and hepatobiliary inflammation, including Crohn’s disease and ulcerative colitis in both human and mouse, and primary sclerosing cholangitis; inflammatory skin disease, including psoriasis and preclinical MC903 and oxazolone mouse models; atherosclerosis; interstitial lung disease; and kidney transplant biology and graft rejection
- Built the biological rationale for a bispecific target by bringing clinical single-cell and spatial transcriptomics into the biomarker strategy. It showed a significant expression correlation with a canonical regulatory T-cell marker before any experimental validation had run
- Built and trained a variational autoencoder to pull strain-level technical variation apart from tumour biomarker signal across genetically diverse mouse strains. It is meant to carry candidate biomarkers over to human population heterogeneity
- Acted as the computational contact point for more than four teams working several projects at once, turning multi-omic results into plain language and helping shape decisions from experimental design through to which validation experiments to run next. Delivered a reusable single-cell and spatial immune atlas and an AWS and Docker HPC environment I built from scratch, both documented so anyone on the team could rerun and build on the work