Interpretation of cancer mutations using a multiscale map of protein systems

How Artificial Intelligence Unravels the Complex Web of Cancer Drug Response

The intersection of precision medicine and artificial intelligence (AI) holds profound implications for cancer treatment, with the potential to significantly advance our understanding of drug responses based on the intricate architecture of tumor cells. A recent study by Park and colleagues titled "A Deep Learning Model of Tumor Cell Architecture Elucidates Response and Resistance to CDK4/6 Inhibitors" epitomizes this intersection by leveraging an interpretable deep learning model grounded in a comprehensive map of multiprotein assemblies in cancer, known as Nested Systems in Tumors. This study not only elucidates mechanisms underlying the response to CDK4/6 inhibitors in breast cancer therapy but also highlights the critical role of model interpretability leading to new mechanistic insights.

Cell Maps for Artificial Intelligence: AI-Ready Maps of Human Cell Architecture from Disease-Relevant Cell Lines

This article describes the Cell Maps for Artificial Intelligence (CM4AI) project and its goals, methods, standards, current datasets, software tools , status, and future directions. CM4AI is the Functional Genomics Data Generation Project in the U.S. National Institute of Health's (NIH) Bridge2AI program. Its overarching mission is to produce ethical, AI-ready datasets of cell architecture, inferred from multimodal data collected for human cell lines, to enable transformative biomedical AI research.

Copy number alterations: a catastrophic orchestration of the breast cancer genome

Breast cancer (BCa) is a prevalent malignancy that predominantly affects women around the world. Somatic copy number alterations (CNAs) are tumor-specific amplifications or deletions of DNA segments that often drive BCa development and therapy resistance. Hence, the complex patterns of CNAs complement BCa classification systems. In addition, understanding the precise contributions of CNAs is essential for tailoring personalized treatment approaches. This review highlights how tumor evolution drives the acquisition of CNAs, which in turn shape the genomic landscapes of BCas. It also discusses advanced methodologies for identifying recurrent CNAs, studying CNAs in BCa and their clinical impact.

Network-based approach elucidates critical genes in BRCA subtypes and chemotherapy response in triple negative breast cancer

Breast cancers (BRCA) exhibit substantial transcriptional heterogeneity, posing a significant clinical challenge. The global transcriptional changes in a disease context, however, are likely mediated by few key genes which reflect disease etiology better than the differentially expressed genes (DEGs). We apply our network-based tool PathExt to 1,059 BRCA tumors across 4 subtypes to identify key mediator genes in each subtype. Compared to conventional differential expression analysis, PathExt-identified genes exhibit greater concordance across tumors, revealing shared and subtype-specific biological processes; better recapitulate BRCA-associated genes in multiple benchmarks, and are more essential in BRCA subtype-specific cell lines. Single-cell transcriptomic analysis reveals a subtype-specific distribution of PathExt-identified genes in multiple cell types from the tumor microenvironment. Application of PathExt to a TNBC chemotherapy response dataset identified subtype-specific key genes and biological processes associated with resistance. We described putative drugs that target key genes potentially mediating drug resistance.

MT-125 Inhibits Non-Muscle Myosin IIA and IIB, Synergizes with Oncogenic Kinase Inhibitors, and Prolongs Survival in Glioblastoma

We have identified a NMIIA and IIB-specific small molecule inhibitor, MT-125, and have studied its effects in GBM. MT-125 has high brain penetrance and retention and an excellent safety profile; blocks GBM invasion and cytokinesis, consistent with the known roles of NMII; and prolongs survival as a single agent in murine GBM models. MT-125 increases signaling along both the PDGFR- and MAPK-driven pathways through a mechanism that involves the upregulation of reactive oxygen species, and it synergizes with FDA-approved PDGFR and mTOR inhibitors in vitro . Combining MT-125 with sunitinib, a PDGFR inhibitor, or paxalisib, a combined PI3 Kinase/mTOR inhibitor significantly improves survival in orthotopic GBM models over either drug alone, and in the case of sunitinib, markedly prolongs survival in ∼40% of mice. Our results provide a powerful rationale for developing NMII targeting strategies to treat cancer and demonstrate that MT-125 has strong clinical potential for the treatment of GBM.

Highlights

MT-125 is a highly specific small molecule inhibitor of non-muscle myosin IIA and IIB, is well-tolerated, and achieves therapeutic concentrations in the brain with systemic dosing.Treating preclinical models of glioblastoma with MT-125 produces durable improvements in survival.MT-125 stimulates PDGFR- and MAPK-driven signaling in glioblastoma and increases dependency on these pathways.Combining MT-125 with an FDA-approved PDGFR inhibitor in a mouse GBM model synergizes to improve median survival over either drug alone, and produces tumor free, prolonged survival in over 40% of mice.

Evaluation of large language models for discovery of gene set function

Gene set analysis is a mainstay of functional genomics, but it relies on curated databases of gene functions that are incomplete. Here we evaluate five Large Language Models (LLMs) for their ability to discover the common biological functions represented by a gene set, substantiated by supporting rationale, citations and a confidence assessment. Benchmarking against canonical gene sets from the Gene Ontology, GPT-4 confidently recovered the curated name or a more general concept (73% of cases), while benchmarking against random gene sets correctly yielded zero confidence. Gemini-Pro and Mixtral-Instruct showed ability in naming but were falsely confident for random sets, whereas Llama2-70b had poor performance overall. In gene sets derived from 'omics data, GPT-4 identified novel functions not reported by classical functional enrichment (32% of cases), which independent review indicated were largely verifiable and not hallucinations. The ability to rapidly synthesize common gene functions positions LLMs as valuable 'omics assistants.

A deep learning model of tumor cell architecture elucidates response and resistance to CDK4/6 inhibitors

Cyclin-dependent kinase 4 and 6 inhibitors (CDK4/6is) have revolutionized breast cancer therapy. However, <50% of patients have an objective response, and nearly all patients develop resistance during therapy. To elucidate the underlying mechanisms, we constructed an interpretable deep learning model of the response to palbociclib, a CDK4/6i, based on a reference map of multiprotein assemblies in cancer. The model identifies eight core assemblies that integrate rare and common alterations across 90 genes to stratify palbociclib-sensitive versus palbociclib-resistant cell lines. Predictions translate to patients and patient-derived xenografts, whereas single-gene biomarkers do not. Most predictive assemblies can be shown by CRISPR-Cas9 genetic disruption to regulate the CDK4/6i response. Validated assemblies relate to cell-cycle control, growth factor signaling and a histone regulatory complex that we show promotes S-phase entry through the activation of the histone modifiers KAT6A and TBL1XR1 and the transcription factor RUNX1. This study enables an integrated assessment of how a tumor's genetic profile modulates CDK4/6i resistance.

Cancer Mutations Converge on a Collection of Protein Assemblies to Predict Resistance to Replication Stress

Rapid proliferation is a hallmark of cancer associated with sensitivity to therapeutics that cause DNA replication stress (RS). Many tumors exhibit drug resistance, however, via molecular pathways that are incompletely understood. Here, we develop an ensemble of predictive models that elucidate how cancer mutations impact the response to common RS-inducing (RSi) agents. The models implement recent advances in deep learning to facilitate multidrug prediction and mechanistic interpretation. Initial studies in tumor cells identify 41 molecular assemblies that integrate alterations in hundreds of genes for accurate drug response prediction. These cover roles in transcription, repair, cell-cycle checkpoints, and growth signaling, of which 30 are shown by loss-of-function genetic screens to regulate drug sensitivity or replication restart. The model translates to cisplatin-treated cervical cancer patients, highlighting an RTK-JAK-STAT assembly governing resistance. This study defines a compendium of mechanisms by which mutations affect therapeutic responses, with implications for precision medicine.

SIGNIFICANCE

Zhao and colleagues use recent advances in machine learning to study the effects of tumor mutations on the response to common therapeutics that cause RS. The resulting predictive models integrate numerous genetic alterations distributed across a constellation of molecular assemblies, facilitating a quantitative and interpretable assessment of drug response. This article is featured in Selected Articles from This Issue, p. 384.

Human-virus protein-protein interactions maps assist in revealing the pathogenesis of viral infection

Many significant viral infections have been recorded in human history, which have caused enormous negative impacts worldwide. Human-virus protein-protein interactions (PPIs) mediate viral infection and immune processes in the host. The identification, quantification, localization, and construction of human-virus PPIs maps are critical prerequisites for understanding the biophysical basis of the viral invasion process and characterising the framework for all protein functions. With the technological revolution and the introduction of artificial intelligence, the human-virus PPIs maps have been expanded rapidly in the past decade and shed light on solving complicated biomedical problems. However, there is still a lack of prospective insight into the field. In this work, we comprehensively review and compare the effectiveness, potential, and limitations of diverse approaches for constructing large-scale PPIs maps in human-virus, including experimental methods based on biophysics and biochemistry, databases of human-virus PPIs, computational methods based on artificial intelligence, and tools for visualising PPIs maps. The work aims to provide a toolbox for researchers, hoping to better assist in deciphering the relationship between humans and viruses.

Desmosome mutations impact the tumor microenvironment to promote melanoma proliferation

Desmosomes are transmembrane protein complexes that contribute to cell-cell adhesion in epithelia and other tissues. Here, we report the discovery of frequent genetic alterations in the desmosome in human cancers, with the strongest signal seen in cutaneous melanoma where desmosomes are mutated in over 70% of cases. In primary but not metastatic melanoma biopsies, the burden of coding mutations on desmosome genes associates with a strong reduction in desmosome gene expression. Analysis by spatial transcriptomics suggests that these expression decreases occur in keratinocytes in the microenvironment rather than in primary melanoma tumor cells. In further support of a microenvironmental origin, we find that loss-of-function knockdowns of the desmosome in keratinocytes yield markedly increased proliferation of adjacent melanocytes in keratinocyte/melanocyte co-cultures. Thus, gradual accumulation of desmosome mutations in neighboring cells may prime melanocytes for neoplastic transformation.

Evaluation of large language models for discovery of gene set function

Gene set analysis is a mainstay of functional genomics, but it relies on manually curated databases of gene functions that are incomplete and unaware of biological context. Here we evaluate the ability of OpenAI's GPT-4, a Large Language Model (LLM), to develop hypotheses about common gene functions from its embedded biomedical knowledge. We created a GPT-4 pipeline to label gene sets with names that summarize their consensus functions, substantiated by analysis text and citations. Benchmarking against named gene sets in the Gene Ontology, GPT-4 generated very similar names in 50% of cases, while in most remaining cases it recovered the name of a more general concept. In gene sets discovered in 'omics data, GPT-4 names were more informative than gene set enrichment, with supporting statements and citations that largely verified in human review. The ability to rapidly synthesize common gene functions positions LLMs as valuable functional genomics assistants.

Pan-cancer proteogenomics connects oncogenic drivers to functional states

Cancer driver events refer to key genetic aberrations that drive oncogenesis; however, their exact molecular mechanisms remain insufficiently understood. Here, our multi-omics pan-cancer analysis uncovers insights into the impacts of cancer drivers by identifying their significant cis-effects and distal trans-effects quantified at the RNA, protein, and phosphoprotein levels. Salient observations include the association of point mutations and copy-number alterations with the rewiring of protein interaction networks, and notably, most cancer genes converge toward similar molecular states denoted by sequence-based kinase activity profiles. A correlation between predicted neoantigen burden and measured T cell infiltration suggests potential vulnerabilities for immunotherapies. Patterns of cancer hallmarks vary by polygenic protein abundance ranging from uniform to heterogeneous. Overall, our work demonstrates the value of comprehensive proteogenomics in understanding the functional states of oncogenic drivers and their links to cancer development, surpassing the limitations of studying individual cancer types.

A multi-scale map of protein assemblies in the DNA damage response

The DNA damage response (DDR) ensures error-free DNA replication and transcription and is disrupted in numerous diseases. An ongoing challenge is to determine the proteins orchestrating DDR and their organization into complexes, including constitutive interactions and those responding to genomic insult. Here, we use multi-conditional network analysis to systematically map DDR assemblies at multiple scales. Affinity purifications of 21 DDR proteins, with/without genotoxin exposure, are combined with multi-omics data to reveal a hierarchical organization of 605 proteins into 109 assemblies. The map captures canonical repair mechanisms and proposes new DDR-associated proteins extending to stress, transport, and chromatin functions. We find that protein assemblies closely align with genetic dependencies in processing specific genotoxins and that proteins in multiple assemblies typically act in multiple genotoxin responses. Follow-up by DDR functional readouts newly implicates 12 assembly members in double-strand-break repair. The DNA damage response assemblies map is available for interactive visualization and query (ccmi.org/ddram/).

Network-based approach elucidates critical genes in BRCA subtypes and chemotherapy response in Triple Negative Breast Cancer

Breast cancers exhibit substantial transcriptional heterogeneity, posing a significant challenge to the prediction of treatment response and prognostication of outcomes. Especially, translation of TNBC subtypes to the clinic remains a work in progress, in part because of a lack of clear transcriptional signatures distinguishing the subtypes. Our recent network-based approach, PathExt, demonstrates that global transcriptional changes in a disease context are likely mediated by a small number of key genes, and these mediators may better reflect functional or translationally relevant heterogeneity. We apply PathExt to 1059 BRCA tumors and 112 healthy control samples across 4 subtypes to identify frequent, key-mediator genes in each BRCA subtype. Compared to conventional differential expression analysis, PathExt-identified genes (1) exhibit greater concordance across tumors, revealing shared as well as BRCA subtype-specific biological processes, (2) better recapitulate BRCA-associated genes in multiple benchmarks, and (3) exhibit greater dependency scores in BRCA subtype-specific cancer cell lines. Single cell transcriptomes of BRCA subtype tumors reveal a subtype-specific distribution of PathExt-identified genes in multiple cell types from the tumor microenvironment. Application of PathExt to a TNBC chemotherapy response dataset identified TNBC subtype-specific key genes and biological processes associated with resistance. We described putative drugs that target top novel genes potentially mediating drug resistance. Overall, PathExt applied to breast cancer refines previous views of gene expression heterogeneity and identifies potential mediators of TNBC subtypes, including potential therapeutic targets.

ProAct: quantifying the differential activity of biological processes in tissues, cells, and user-defined contexts

The distinct functions and phenotypes of human tissues and cells derive from the activity of biological processes that varies in a context-dependent manner. Here, we present the Process Activity (ProAct) webserver that estimates the preferential activity of biological processes in tissues, cells, and other contexts. Users can upload a differential gene expression matrix measured across contexts or cells, or use a built-in matrix of differential gene expression in 34 human tissues. Per context, ProAct associates gene ontology (GO) biological processes with estimated preferential activity scores, which are inferred from the input matrix. ProAct visualizes these scores across processes, contexts, and process-associated genes. ProAct also offers potential cell-type annotations for cell subsets, by inferring them from the preferential activity of 2001 cell-type-specific processes. Thus, ProAct output can highlight the distinct functions of tissues and cell types in various contexts, and can enhance cell-type annotation efforts. The ProAct webserver is available at https://netbio.bgu.ac.il/ProAct/.

m6A readers, writers, erasers, and the m6A epitranscriptome in breast cancer

Epitranscriptomic modification of RNA regulates human development, health, and disease. The true diversity of the transcriptome in breast cancer including chemical modification of transcribed RNA (epitranscriptomics) is not well understood due to limitations of technology and bioinformatic analysis. N-6-methyladenosine (m6A) is the most abundant epitranscriptomic modification of mRNA and regulates splicing, stability, translation, and intracellular localization of transcripts depending on m6A association with reader RNA-binding proteins. m6A methylation is catalyzed by the METTL3 complex and removed by specific m6A demethylase ALKBH5, with the role of FTO as an 'eraser' uncertain. In this review, we provide an overview of epitranscriptomics related to mRNA and focus on m6A in mRNA and its detection. We summarize current knowledge on altered levels of writers, readers, and erasers of m6A and their roles in breast cancer and their association with prognosis. We summarize studies identifying m6A peaks and sites in genes in breast cancer cells.

From multi-omics data to the cancer druggable gene discovery: a novel machine learning-based approach

The development of targeted drugs allows precision medicine in cancer treatment and optimal targeted therapies. Accurate identification of cancer druggable genes helps strengthen the understanding of targeted cancer therapy and promotes precise cancer treatment. However, rare cancer-druggable genes have been found due to the multi-omics data's diversity and complexity. This study proposes deep forest for cancer druggable genes discovery (DF-CAGE), a novel machine learning-based method for cancer-druggable gene discovery. DF-CAGE integrated the somatic mutations, copy number variants, DNA methylation and RNA-Seq data across ˜10 000 TCGA profiles to identify the landscape of the cancer-druggable genes. We found that DF-CAGE discovers the commonalities of currently known cancer-druggable genes from the perspective of multi-omics data and achieved excellent performance on OncoKB, Target and Drugbank data sets. Among the ˜20 000 protein-coding genes, DF-CAGE pinpointed 465 potential cancer-druggable genes. We found that the candidate cancer druggable genes (CDG) are clinically meaningful and divided the CDG into known, reliable and potential gene sets. Finally, we analyzed the omics data's contribution to identifying druggable genes. We found that DF-CAGE reports druggable genes mainly based on the copy number variations (CNVs) data, the gene rearrangements and the mutation rates in the population. These findings may enlighten the future study and development of new drugs.

Where protein structure and cell diversity meet

Protein-protein interaction networks - interactomes - are charted with the hope to understand how phenotypes emerge and how they are altered in disease states. Early efforts to map interactomes have focused on the assembly of context agnostic, reference networks. However, recent studies have mapped interactomes across different cell lines and tissues, finding highly variable interactomes due to the rewiring of protein-protein interactions in different contexts. Increasing evidence points to significant links between protein structure and interactome diversity seen across cell types and tissues. We discuss how recent findings support the key role of alternative splicing and phosphorylation, two well-established regulators of protein structural and functional diversity, in defining cell type- and tissue-specific interactomes. Moreover, we show that intrinsically disordered protein regions are most favorably equipped to support interactome rewiring by acting as hubs of protein structure and function regulation.

Multi-omics peripheral and core regions of cancer

Thousands of genes are perturbed by cancer, and these disturbances can be seen in transcriptome, methylation, somatic mutation, and copy number variation omics studies. Understanding their connectivity patterns as an omnigenic neighbourhood in a molecular interaction network (interactome) is a key step towards advancing knowledge of the molecular mechanisms underlying cancers. Here, we introduce a unified connectivity line (CLine) to pinpoint omics-specific omnigenic patterns across 15 curated cancers. Taking advantage of the universality of CLine, we distinguish the peripheral and core genes for each omics aspect. We propose a network-based framework, multi-omics periphery and core (MOPC), to combine peripheral and core genes from different omics into a button-like structure. On the basis of network proximity, we provide evidence that core genes tend to be specifically perturbed in one omics, but the peripheral genes are diversely perturbed in multiple omics. And the core of one omics is regulated by multiple omics peripheries. Finally, we take the MOPC as an omnigenic neighbourhood, describe its characteristics, and explore its relative contribution to network-based mechanisms of cancer. We were able to present how multi-omics perturbations percolate through the human interactome and contribute to an integrated periphery and core.

CORUM: the comprehensive resource of mammalian protein complexes-2022

The CORUM database has been providing comprehensive reference information about experimentally characterized, mammalian protein complexes and their associated biological and biomedical properties since 2007. Given that most catalytic and regulatory functions of the cell are carried out by protein complexes, their composition and characterization is of greatest importance in basic and disease biology. The new CORUM 4.0 release encompasses 5204 protein complexes offering the largest and most comprehensive publicly available dataset of manually curated mammalian protein complexes. The CORUM dataset is built from 5299 different genes, representing 26% of the protein coding genes in humans. Complex information from 3354 scientific articles is mainly obtained from human (70%), mouse (16%) and rat (9%) cells and tissues. Recent curation work includes sets of protein complexes, Functional Complex Groups, that offer comprehensive collections of published data in specific biological processes and molecular functions. In addition, a new graphical analysis tool was implemented that displays co-expression data from the subunits of protein complexes. CORUM is freely accessible at http://mips.helmholtz-muenchen.de/corum/.

Big data in basic and translational cancer research

Historically, the primary focus of cancer research has been molecular and clinical studies of a few essential pathways and genes. Recent years have seen the rapid accumulation of large-scale cancer omics data catalysed by breakthroughs in high-throughput technologies. This fast data growth has given rise to an evolving concept of 'big data' in cancer, whose analysis demands large computational resources and can potentially bring novel insights into essential questions. Indeed, the combination of big data, bioinformatics and artificial intelligence has led to notable advances in our basic understanding of cancer biology and to translational advancements. Further advances will require a concerted effort among data scientists, clinicians, biologists and policymakers. Here, we review the current state of the art and future challenges for harnessing big data to advance cancer research and treatment.

A network view of human immune system and virus-human interaction

The immune system is highly networked and complex, which is continuously changing as encountering old and new pathogens. However, reductionism-based researches do not give a systematic understanding of the molecular mechanism of the immune response and viral pathogenesis. Here, we present HUMPPI-2022, a high-quality human protein-protein interaction (PPI) network, containing > 11,000 protein-coding genes with > 78,000 interactions. The network topology and functional characteristics analyses of the immune-related genes (IRGs) reveal that IRGs are mostly located in the center of the network and link genes of diverse biological processes, which may reflect the gene pleiotropy phenomenon. Moreover, the virus-human interactions reveal that pan-viral targets are mostly hubs, located in the center of the network and enriched in fundamental biological processes, but not for coronavirus. Finally, gene age effect was analyzed from the view of the host network for IRGs and virally-targeted genes (VTGs) during evolution, with IRGs gradually became hubs and integrated into host network through bridging functionally differentiated modules. Briefly, HUMPPI-2022 serves as a valuable resource for gaining a better understanding of the composition and evolution of human immune system, as well as the pathogenesis of viruses.

From systems to structure - using genetic data to model protein structures

Understanding the effects of genetic variation is a fundamental problem in biology that requires methods to analyse both physical and functional consequences of sequence changes at systems-wide and mechanistic scales. To achieve a systems view, protein interaction networks map which proteins physically interact, while genetic interaction networks inform on the phenotypic consequences of perturbing these protein interactions. Until recently, understanding the molecular mechanisms that underlie these interactions often required biophysical methods to determine the structures of the proteins involved. The past decade has seen the emergence of new approaches based on coevolution, deep mutational scanning and genome-scale genetic or chemical-genetic interaction mapping that enable modelling of the structures of individual proteins or protein complexes. Here, we review the emerging use of large-scale genetic datasets and deep learning approaches to model protein structures and their interactions, and discuss the integration of structural data from different sources.

A protein network map of head and neck cancer reveals PIK3CA mutant drug sensitivity

[Figure: see text].

Identifying cancer drivers

[Figure: see text].

A protein interaction landscape of breast cancer

[Figure: see text].