SCOUT

Matched tumour–normal whole-genome sequencing (WGS) datasets encompassing multiple tumour types, multiregion and longitudinal sampling, and pre- and post-treatment conditions.

Comprehensive characterization of seven tumour types, including five solid cancers and two haematological malignancies.

Synthetic 150 bp paired-end FASTQ reads simulating Illumina NovaSeq 6000 sequencing, with realistic base-calling quality profiles.

Systematic benchmarking of somatic and germline variant callers, copy number alteration methods, tumour purity estimators, and evolutionary inference frameworks for subclonal reconstruction and phylogenetic modelling.

Simulated tumour–normal datasets across different tumour purities (0.3, 0.6, and 0.9) and sequencing depths (50x, 100x, and 150x), reflecting common WGS conditions encountered in cancer research.

Simulations are grounded in real human genetic diversity using germline polymorphisms derived from the 1000 Genomes Project, ensuring realistic population-level background variation in all tumour–normal samples.

Tumour-specific driver alterations are informed by the Cancer Gene Census, enabling realistic oncogenic architectures that reflect known cancer pathways and disease-specific mutational landscapes.

Mutational processes are modelled using COSMIC signatures, capturing context-dependent substitution patterns and indel formation mechanisms that reproduce tumour-type-specific and treatment-related mutational spectra.

Copy-number alterations are derived from PCAWG and Hartwig datasets, reproducing realistic patterns of aneuploidy, focal amplifications, and deletions across diverse cancer types.

SCOUT integrates clonal expansion, spatial growth, and treatment-driven selection into a unified framework, enabling coherent evolution across cellular, genomic, and population scales.

Genomic signals such as variant allele frequencies, copy-number states, mutational signatures, and phylogenetic structure arise jointly from shared evolutionary processes, producing correlated patterns across multiple molecular layers.

SCOUT replaces simplified spike-in methods with explicit evolutionary ground truth, capturing tumour growth, subclonal diversification, and treatment-driven selection in a unified simulation framework.

Methods are assessed against explicit evolutionary mechanisms, including clonal ancestry, spatial structure, mutational processes, and treatment selection, enabling direct comparison between inferred outputs and true tumour histories.

Extension of benchmarking to single-cell DNA sequencing technologies This will enable evaluation of SNV and copy-number calling performance, as well as phylogenetic reconstruction accuracy at single-cell resolution under realistic noise and coverage constraints.

Integration of phenotypic evolution with genomic changes through coupled simulation of epigenetic regulation and transcriptional programs, enabling joint modelling of genotype–phenotype interplay during tumour progression.

New simulated experimental layers including single-cell RNA-seq and single-cell ATAC-seq, capturing gene expression dynamics and chromatin accessibility landscapes across evolving tumour subclones.

We generated SCOUT using ProCESS (Programmable Cancer Evolution Spatial Simulator), a framework that integrates spatial tumour evolution with synthetic whole-genome sequencing.

We benchmarked somatic and germline variant calling and copy-number detection in tumour–normal WGS using nf-core/sarek, evaluating performance across tumour purity, sequencing depth, and genomic profiles against ProCESS-derived ground truth.

We benchmarked tumour evolution inference methods using nf-core/tumourevo, evaluating driver mutation detection, subclonal deconvolution, and mutational signatures analysis. Using ProCESS-generated ground-truth evolutionary scenarios, we assessed the accuracy of reconstructed clonal architectures and mutational processes.

An R package providing streamlined access to SCOUT datasets generated via ProCESS. It enables efficient download and querying of processed sequencing data, along with convenient functions to access associated evolutionary ground truth and benchmarking results.

A public GitHub repository containing all scripts required to reproduce the analyses presented in the SCOUT study, including data processing workflows, benchmarking procedures, and figure generation.

A containerised computing environment providing full reproducibility of the SCOUT analysis pipeline. The Docker image includes all dependencies, tools, and workflow configurations required to regenerate the analyses and figures.

Paired-end FASTQ files for each sample and purity level at the maximum simulated coverage are available in ENA under the accession number PRJEB97253T.

Simulated ground truth mutations from ProCESS, somatic SNVs and CNA calls from nf-core/sarek and output files from nf-core/tumourevo are available in Zenodo.