The rapid advancement of high-throughput sequencing and other assay technologies has resulted in the generation of large and complex multi-omics datasets, offering unprecedented opportunities for advancing precision medicine. However, multi-omics data integration remains challenging due to the high-dimensionality, heterogeneity, and frequency of missing values across data types. Computational methods leveraging statistical and machine learning approaches have been developed to address these issues and uncover complex biological patterns, improving our understanding of disease mechanisms. Here, we comprehensively review state-of-the-art multi-omics integration methods with a focus on deep generative models, particularly variational autoencoders (VAEs) that have been widely used for data imputation, augmentation, and batch effect correction. We explore the technical aspects of VAE loss functions and regularisation techniques, including adversarial training, disentanglement, and contrastive learning. Moreover, we highlight recent advancements in foundation models and multimodal data integration, outlining future directions in precision medicine research.

Reproducible research is the bedrock of experimental science. To enable the deployment of large-scale proteomics, we assess the reproducibility of mass spectrometry (MS) over time and across instruments and develop computational methods for improving quantitative accuracy. We perform 1560 data independent acquisition (DIA)-MS runs of eight samples containing known proportions of ovarian and prostate cancer tissue and yeast, or control HEK293T cells. Replicates are run on six mass spectrometers operating continuously with varying maintenance schedules over four months, interspersed with ~5000 other runs. We utilise negative controls and replicates to remove unwanted variation and enhance biological signal, outperforming existing methods. We also design a method for reducing missing values. Integrating these computational modules into a pipeline (ProNorM), we mitigate variation among instruments over time and accurately predict tissue proportions. We demonstrate how to improve the quantitative analysis of large-scale DIA-MS data, providing a pathway toward clinical proteomics.

MYCN gene amplification in neuroblastoma drives a gene expression program that correlates strongly with aggressive disease. Mechanistically, trimethylation of histone H3 lysine 4 (H3K4) at target gene promoters is a strict prerequisite for this transcriptional program to be enacted. WDR5 is a histone H3K4 presenter that has been found to have an essential role in H3K4 trimethylation. For this reason, in this study, we investigated the relationship between WDR5-mediated H3K4 trimethylation and N-Myc transcriptional programs in neuroblastoma cells. N-Myc upregulated WDR5 expression in neuroblastoma cells. Gene expression analysis revealed that WDR5 target genes included those with MYC-binding elements at promoters such as MDM2. We showed that WDR5 could form a protein complex at the MDM2 promoter with N-Myc, but not p53, leading to histone H3K4 trimethylation and activation of MDM2 transcription. RNAi-mediated attenuation of WDR5 upregulated expression of wild-type but not mutant p53, an effect associated with growth inhibition and apoptosis. Similarly, a small-molecule antagonist of WDR5 reduced N-Myc/WDR5 complex formation, N-Myc target gene expression, and cell growth in neuroblastoma cells. In MYCN-transgenic mice, WDR5 was overexpressed in precancerous ganglion and neuroblastoma cells compared with normal ganglion cells. Clinically, elevated levels of WDR5 in neuroblastoma specimens were an independent predictor of poor overall survival. Overall, our results identify WDR5 as a key cofactor for N-Myc–regulated transcriptional activation and tumorigenesis and as a novel therapeutic target for MYCN-amplified neuroblastomas. Cancer Res; 75(23); 5143–54. ©2015 AACR.

Methylated cytosines (5mCs) are frequently mutated in the genome. However, no studies have yet comprehensively analysed mutation–methylation associations across cancer types. Here we analyse 916 cancer genomes, together with tissue type-specific methylation and replication timing data. We describe a strong mutation–methylation association across colorectal cancer subtypes, most interestingly in samples with microsatellite instability (MSI) or Polymerase epsilon (POLE) exonuclease domain mutations. By analysing genomic regions with differential mismatch repair (MMR) efficiency, we suggest a possible role for MMR in the correction of 5mC deamination events, potentially accounting for the high rate of 5mC mutation accumulation in MSI tumours. Additionally, we propose that mutant POLE asserts a mutator phenotype specifically at 5mCs, and we find coding mutation hotspots in POLE-mutant cancers at highly-methylated CpGs in the tumour-suppressor genes APC and TP53. Finally, using multivariable regression models, we demonstrate that different cancers exhibit distinct mutation–methylation associations, with DNA repair influencing such associations in certain cancer genomes. Taken together, we find differential associations with methylation that are vital for accurately predicting expected mutation loads across cancer types. Our findings reveal links between methylation and common mutation and repair processes, with these mechanisms defining a key part of the mutational landscape of cancer genomes.

CTCF binding sites are frequently mutated in cancer, but how these mutations accumulate and whether they broadly perturb CTCF binding are not well understood. Here, we report that skin cancers exhibit a highly specific asymmetric mutation pattern within CTCF motifs attributable to ultraviolet irradiation and differential nucleotide excision repair (NER). CTCF binding site mutations form independently of replication timing and are enriched at sites of CTCF/cohesin complex binding, suggesting a role for cohesin in stabilizing CTCF-DNA binding and impairing NER. Performing CTCF ChIP-seq in a melanoma cell line, we show CTCF binding site mutations to be functional by demonstrating allele-specific reduction of CTCF binding to mutant alleles. While topologically associating domains with mutated CTCF anchors in melanoma contain differentially expressed cancer-associated genes, CTCF motif mutations appear generally under neutral selection. However, the frequency and potential functional impact of such mutations in melanoma highlights the need to consider their impact on cellular phenotype in individual genomes.