Kieran R. Campbell

Multi-view deep learning of highly multiplexed imaging data improves association of cell states with clinical outcomes

This study uses multi-modal VAEs to integrate multiple views of highly multiplexed imaging data, improving the prediction of clinical outcomes and identification of cell states.

Computational design and evaluation of optimal bait sets for scalable proximity proteomics

The paper develops GENBAIT, a genetic algorithm-based tool to optimize bait selection for BioID experiments, reducing bait numbers while maintaining dataset coverage and structure.

Rakaia: interactive discovery of spatial biology at scale

Rakaia enables interactive exploration of spatial biology data through a browser-based platform, identifying cell types in non-malignant human breast tissue.

Integrated spatial proteomics of human PDAC uncovers an expanded tumour-immune-stroma spectrum with genomic associations

This study integrates spatial proteomics and genomics in PDAC to identify tumour-immune-stroma interactions and their genomic associations, showing multi-omics models improve patient survival prediction.

Damage and Misrepair Signatures: Compact Representations of Pan-cancer Mutational Processes

The paper introduces DAMUTA, a model that distinguishes DNA damage and misrepair processes in mutational signatures, showing improved resolution over existing methods in pan-cancer analysis.

Automated registration of spatial expression data scales multimodal integration to large cohorts

The paper introduces a Bayesian optimization framework for automated registration of spatial expression data across modalities, achieving higher success rates on a large dataset of 954 image pairs.

Learning copy number dependent variation in single tumour cell transcriptomes with deep generative models

The paper introduces ISOMERIC, a deep learning framework to separate copy number-dependent and independent variation in single-cell transcriptomes, showing distinct associations with clinical subtypes and tumour microenvironment phenotypes.

Multi-modal Disentanglement of Spatial Transcriptomics and Histopathology Imaging

This paper introduces SpatialDIVA, a method to disentangle spatial transcriptomics and histopathology imaging data, showing it outperforms baselines in variation separation and batch effect removal.

Segmentation aware probabilistic phenotyping of single-cell spatial protein expression data

STARLING quantifies cell populations from spatial protein expression data while accounting for segmentation errors, identifying known cell types and quantifying heterogeneity in human tonsil tissue.

Interactive design and validation of antibody panels using single-cell RNA-seq atlases

The paper introduces Cytomarker, an interactive platform for designing antibody panels using single-cell RNA-seq data, and validates it on human mammary tissue subpopulations.

Inferring tumour microenvironment ecosystems from scRNA-seq atlases

The paper infers multicellular ecosystems from nonspatial scRNA-seq data and shows they predict therapy response and are linked to somatic mutations.

Single cell decoding of drug induced transcriptomic reprogramming in triple negative breast cancers

The paper identifies genomic and non-genomic contributions to drug resistance in triple negative breast cancer using single cell transcriptomics and copy number data.

Spatial proteomic mapping of human nuclear bodies reveals new functional insights into RNA regulation

The paper maps human nuclear bodies using proximity proteomics, identifying new components and regulatory functions in RNA regulation.

Beyond benchmarking: towards predictive models of dataset-specific single-cell RNA-seq pipeline performance

The paper applies 288 scRNA-seq pipelines to 86 datasets and builds machine learning models to predict pipeline performance based on dataset and pipeline characteristics.

The impacts of active and self-supervised learning on efficient annotation of single-cell expression data

This paper benchmarks active and self-supervised learning strategies for single-cell data annotation, showing benefits in handling cell type imbalance and demonstrating marker-aware heuristics improve accuracy.

Cancer phylogenetic tree inference at scale from 1000s of single cell genomes

The paper introduces a scalable Bayesian method for phylogenetic tree inference from single-cell whole genome sequencing data, demonstrating its application on three datasets.

The differential impacts of dataset imbalance in single-cell data integration

This study evaluates how dataset imbalance affects single-cell RNA sequencing integration and introduces metrics to assess integration stability in imbalanced settings.

Multi-objective Bayesian Optimization with Heuristic Objectives for Biomedical and Molecular Data Analysis Workflows

The paper introduces a multi-objective Bayesian optimization method that adaptively updates scalarization functions for biomedical data analysis workflows, demonstrating it on single-cell RNA sequencing and imaging data.

Automated assignment of cell identity from single-cell multiplexed imaging and proteomic data

Astir assigns cell identities to single-cell imaging and proteomic data by integrating prior knowledge, outperforming existing methods in accuracy and scalability.

Dissociation of solid tumour tissues with cold active protease for single-cell RNA-seq minimizes conserved collagenase associated stress responses

This study identifies transcriptional stress responses in single-cell RNA-seq data from dissociated tumour tissues, showing that collagenase digestion induces a conserved stress response minimized by cold active protease.

clonealign: statistical integration of independent single-cell RNA & DNA-seq from human cancers

clonealign integrates single-cell RNA and DNA sequencing data from cancer samples to identify clone-specific gene expression patterns and dysregulated pathways.

Probabilistic cell type assignment of single-cell transcriptomic data reveals spatiotemporal microenvironment dynamics in human cancers

CellAssign probabilistically assigns cell types to scRNA-seq data, revealing spatiotemporal dynamics in human cancers like ovarian and follicular lymphoma.

A descriptive marker gene approach to single-cell pseudotime inference

The paper introduces Ouija, a method that infers pseudotime from marker genes to understand cell trajectory dynamics and detect metastable states.

Uncovering genomic trajectories with heterogeneous genetic and environmental backgrounds across single-cells and populations

The paper introduces a statistical framework to model pseudotime trajectories in heterogeneous genetic and environmental backgrounds, demonstrating its application to single-cell and population-level data.