Weischenfeldt Group

We combine computational and experimental approaches to understand how cancers evolve. We study the mutational processes and clonal evolution trajectories that shape tumour genomes, and how these changes drive metastasis and treatment resistance.

Working with clinicians and research partners in Denmark and internationally, we aim to turn our findings into biomarkers that identify patients with aggressive disease and guide personalised treatment.

Vision & research questions

Cancer is a disease of the genome, fuelled by the accumulation of genomic alterations that improve the fitness of the cell. Our overarching goal is to understand how tumour cells escape and evolve in time and space to cause metastatic, treatment-resistant disease – and to use this knowledge to improve risk stratification and treatment selection for patients.

We combine computational cancer genomics, whole-genome sequencing of large patient cohorts, and single-cell and spatial technologies to address the following questions:

Which mutational processes shape cancer genomes, and can their “fingerprints” predict which tumours will become aggressive?
How do complex structural variants, extrachromosomal DNA (ecDNA) and changes to the 3D genome activate oncogenes and inactivate tumour suppressor genes?
How do genetic and epigenetic “second hits” inactivate tumour suppressor genes that are missed by conventional analyses?
How do clonal evolution and cellular plasticity interact to drive metastatic dissemination and therapy resistance?
How can genome-informed biomarkers be brought into the clinic to tailor treatment to the individual patient?

Main findings

ntegrated mutational footprints and genomic drivers define aggressive prostate cancer

In the largest whole-genome study of prostate cancer to date (Pan Prostate Cancer Group, ~1,000 patients), we integrated substitution, insertion–deletion, copy-number and complex structural variant signatures and identified eight integrated mutational footprints (IMFs) that explain the mutational processes in 85% of primary prostate cancers. Four footprints, present in 37% of tumours, were associated with shorter time to metastasis, and the footprints predicted response to androgen receptor pathway inhibitors. A companion study mapped the full spectrum of driver alterations, found at least one recurrent alteration in 97% of tumours and defined four genomic “oncotypes”, two of which carry a poor prognosis.

Gruber, Olsen, Hernando, Cheng et al., Nature, 2026 (PMID: 42717097)
Pellegrina, Kiriy, Cheng et al., Nature Communications, 2026 (PMID: 42888190)
Gerhäuser et al., Cancer Cell, 2018 (PMID: 30537516)
Weischenfeldt et al., Cancer Cell, 2013 (PMID: 23410972)

Clonal evolution and cellular plasticity drive metastasis and treatment resistance

Using whole-genome and single-nucleus RNA sequencing of spatially distinct tumour samples, we showed that prostate cancer metastasises through both single-clone and multi-clone seeding, that the primary tumour continues to evolve after metastatic spread, and that metastatic seeding converges on androgen receptor-independent transcriptional states. In glioblastoma, we developed methods to resolve ecDNA at single-cell resolution, showing that rare ecDNA-carrying clones can expand to dominate relapse, and showed that immunotherapy drives a shift towards mesenchymal tumour cell states.

Mikutenaite, Karadoulama et al., Nature Communications, 2025 (PMID: 41436444)
Hendriksen et al., Genome Biology, 2026 (PMID: 41606654)
Hendriksen et al., Neuro-Oncology, 2024 (PMID: 38695342)
Gerhäuser et al., Cancer Cell, 2018 (PMID: 30537516)

New mechanisms of tumour suppressor gene inactivation

We developed Epi2Hit, which combines whole genomes, epigenomes and gene expression to detect biallelic inactivation through DNA methylation of one allele and loss of the other. Applied to more than a thousand cancers, this revealed epigenetic inactivation of ZFHX3 at a frequency similar to TP53, linked to MYC activation and shorter time to metastasis. An enzyme family-centred pan-cancer analysis identified helicases as frequently mutated cancer genes and hemizygous loss as a common tumour suppression mechanism.

Kiriy et al., Nature Communications, 2026 (PMID: 42716948)
Voßgröne, Favero et al., Science Advances, 2026 (PMID: 41719398)
Flynn et al., Molecular Systems Biology, 2024 (PMID: 39026103)
Tulstrup et al., Nature Communications, 2021 (PMID: 34663818)

Structural variants, the 3D genome and enhancer hijacking

We have shown how structural variants are guided by, and in turn reshape, the 3D organisation of the genome, and how they activate oncogenes through enhancer hijacking – including SV-mediated hijacking of onco-enhancers in cancer and, beyond cancer, a non-coding enhancer variant that causes an inherited cardiac syndrome.

Pellegrina, Kiriy, Cheng et al., Nature Communications, 2026 (PMID: 42888190)
Dubois, Sidiropoulos et al., Nature Reviews Cancer, 2022 (PMID: 35764888)
Sidiropoulos et al., Genome Research, 2022 (PMID: 35177558)
Christensen et al., European Heart Journal, 2025 (PMID: 40208226)
Li et al., Nature, 2020 (PMID: 32025012)
Rheinbay et al., Nature, 2020 (PMID: 32025015)
Weischenfeldt et al., Nature Genetics, 2017 (PMID: 27869826)

Main projects

Mutational processes and genomic biomarkers for risk stratification in prostate cancer

We use whole-genome sequencing and machine learning to define mutational processes, driver alterations and genomic subtypes, and to develop biomarkers that distinguish indolent from aggressive disease and predict treatment response.

Clonal evolution, plasticity and metastatic dissemination

We reconstruct the clonal evolution of tumours from multi-region, single-cell and spatial data to understand how clonal dynamics, transcriptional plasticity and the tumour microenvironment interact to drive metastasis and treatment resistance in prostate cancer and glioblastoma.

Complex structural variants, ecDNA and the 3D genome

We study how complex structural variants and extrachromosomal DNA emerge, how they alter chromatin looping and gene regulation (e.g. enhancer hijacking), and how they contribute to oncogene amplification and tumour heterogeneity. We integrate somatic alterations with epigenetic data and chromatin conformation methods such as Hi-C.

Epigenetic and genetic second hits in tumour suppressor genes

We develop integrative computational methods to identify tumour suppressor genes inactivated by combinations of genomic loss, DNA methylation and other mechanisms that escape standard mutation-based analyses.

Technology

We develop and apply computational and molecular biology approaches. We have optimised and integrated a broad range of sequencing technologies on clinical material, including whole-genome sequencing, RNA-seq, ATAC-seq, Hi-C and Capture-Hi-C, DNA methylation profiling, targeted panel sequencing, single-cell and single-nucleus RNA-seq, single-cell multiome, spatial transcriptomics and long-read sequencing.

We develop open computational methods – for example for integrated mutational signatures, biallelic epigenetic inactivation (Epi2Hit) and single-cell ecDNA detection (ecSingle) – and apply machine learning to large cancer genome cohorts. Analyses run on high-performance computing systems; we use Git for version control and containers (Docker) for portable, reproducible pipelines. Code is shared on GitHub.