SEARCH · Search bioRxiv

Results for “Cancer Biology”

Search indexed bioRxiv preprints in genomics, neuroscience, cell biology and bioinformatics. Read source abstracts and check manuscript versions; preprints are not peer reviewed.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 181 records · Page 10Linked to original sources

bioRxiv2023

A Rapid, Functional sgRNA Screening Method for Generating Murine RET and NTRK1 Fusion Oncogenes

CRISPR/Cas9 gene editing technology is an indispensable and powerful tool in the field of cancer biology. To conduct successful CRISPR-based experiments, it is crucial that sgRNAs generate their designed alterations. Here, we describe a simple and efficient sgRNA screening method for validating sgRNAs that generate oncogenic gene rearrangements. We used IL3-independence in Ba/F3 cells as an assay to identify sgRNA pairs that generate fusion oncogenes involving the Ret and Ntrk1 tyrosine kinases. We confirmed these rearrangements with PCR or RT-PCR as well as sequencing. Ba/F3 cells harboring Ret or Ntrk1 rearrangements acquired sensitivity to RET and TRK inhibitors, respectively. Adenoviruses encoding Cas9 and sgRNAs that catalyze the Kif5b-Ret and Trim24-Ret rearrangements were intratracheally instilled into mice and yielded lung adenocarcinomas. A cell line (TR.1) was established from a Trim24-Ret positive tumor that exhibited high in vitro sensitivity to RET-specific TKIs. Moreover, orthotopic transplantation of TR.1 cells into the left lung yielded well-defined tumors that shrank in response to LOXO-292 treatment. The method offers an efficient means to validate sgRNAs that successfully target their intended loci for the generation of novel murine oncogene-driven tumor models.

cancer biology↗

bioRxiv2023

Second-order effects of chemotherapy pharmacodynamics and pharmacokinetics on tumor regression and cachexia

Drug dose response curves are ubiquitous in cancer biology, but these curves are often used to measure differential response in first-order effects: the effectiveness of increasing the cumulative dose delivered. In contrast, second-order effects (the variance of drug dose) are often ignored. Knowledge of second-order effects may improve the design of chemotherapy scheduling protocols, leading to improvements in tumor response without changing the total dose delivered. By considering treatment schedules with identical cumulative dose delivered, we optimize treatment by comparing high variance schedules (e.g. high dose, low dose) with low variance schedules (constant dose). We extend a previous framework used to quantify second-order effects, known as antifragility theory, to investigate the role of drug pharmacokinetics. Using a simple one-compartment model, we find that high variance schedules are effective for a wide range of cumulative dose values. Next, using a mouse-parameterized two-compartment model of 5-fluorouracil, we show that the optimal schedule depends on initial tumor volume. Finally, we illustrate the trade-off between tumor response and lean mass preservation. Mathematical modeling indicates that high variance dose schedules provide a potential path forward in mitigating the risk of chemotherapy-associated cachexia by preserving lean mass without sacrificing tumor response.

cancer biology↗

bioRxiv2024

Reliable RNA-seq analysis from FFPE specimens as a means to accelerate cancer-related health disparities research

Whole transcriptome sequencing (WTS/ RNA-Seq) is a ubiquitous tool for investigating cancer biology. RNA isolated from frozen sources limits possible studies for analysis of associations with phenotypes or clinical variables requiring long-term follow-up. Although good correlations are reported in RNA-Seq data from paired frozen and formalin fixed paraffin embedded (FFPE) samples, uncertainties regarding RNA quality, methods of extraction, and data reliability are hurdles to utilization of archival samples. We compared three different platforms for performing RNA-seq using archival FFPE oropharyngeal squamous carcinoma (OPSCC) specimens stored up to 20 years, as part of an investigation of transcriptional profiles related to health disparities. We developed guidelines to purify DNA and RNA from FFPE tissue and perform downstream RNA-seq and DNA SNP arrays. RNA was extracted from 150 specimens, with an average yield of 401.8 ng/cm2 of tissue. Most samples yielded sufficient RNA reads >13,000 protein coding genes which could be used to differentiate HPV-associated from HPV-independent OPSCCs. Co-isolated DNA was used to identify patient ancestry. Utilizing the methods described in this study provides a robust, reliable, and standardized means of DNA & RNA extraction from FFPE as well as a means by which to assure the quality of the data generated.

cancer biology↗

bioRxiv2024

An extra-glycolytic function for hexokinase 2 as an RNA-binding protein regulating SOX10 mRNA translation in melanoma

Several studies have reported the importance of aerobic glycolysis in melanoma development. Although metabolic benefits of glycolysis have been extensively described in tumor cells, the extra-metabolic functions linked to this energetic pathway in melanoma growth and proliferation have not been clearly established yet. Recently, some key glycolytic enzymes, such as GAPDH and PKM2, were reported to regulate mRNA translation. Translational control of gene expression is considered as a critical effector in cancer biology, representing a highly promising area of research. Here, we report that Hexokinase 2 (HK2), a glucose kinase that catalyzes the first step of glycolysis, is an RNA binding protein (RBP) that regulates mRNA translation in melanoma. We show that siRNA-mediated HK2 depletion changes the translational landscape of melanoma cells. Polysome profiling experiments and RNA-Seq indicate that the translational regulation exerted by HK2 is partly independent of the metabolic status or the glycolytic pathway. We found that HK2 specifically regulates the translation of the mRNA encoding SOX10, a transcription factor implicated in the regulation of tumor initiation, maintenance and progression in melanoma. RNA-protein interaction assays, including crosslinking immunoprecipitation (CLIP), indicate that HK2 is an RBP whose interaction with RNA is independent of its hexokinase activity or subcellular localization. We also show that HK2 specifically associates with the 5 untranslated region (5UTR) of the SOX10 mRNA, and that several deletions in this region decreases both HK2-SOX10 mRNA association and SOX10 5 UTR-mediated translation. We further show that HK2-dependent SOX10 translational regulation is involved in melanoma cell proliferation and colony formation. Collectively, our data highlight a non-metabolic function of HK2, indicating that melanoma cells may enhance glycolysis for purposes beyond simple anabolism.

cancer biology↗

bioRxiv2024

An integrative RNA spliceosomic landscape of pancreatic neuroendocrine tumors unveils novel clinicomolecular associations

Alterations in alternative splicing are emerging as a novel hallmark in cancer biology, offering new insights. However, integrative analyses of splicing are still scarce, particularly in rare cancers such as pancreatic neuroendocrine tumors (PanNETs). These tumors are highly heterogeneous, complicating diagnosis and treatment. This study is the first to comprehensively investigate the RNA splicing landscape in PanNETs, identifying distinct spliceosomic profiles correlated with unique clinical and molecular characteristics. We analyzed RNA-seq data from 174 samples, identifying three distinct spliceosomic groups (SPN1, SPN2, SPN3) with unique clinical and molecular characteristics. SPN1 exhibited intermediate clinical features and specific splicing machinery profile, SPN2 was associated with frequent mutations in MEN1 and DAXX/ATRX genes, and SPN3 showed a prevalence of well-differentiated tumors with distinct splicing patterns. These groups were linked to different clinical outcomes and activated key biological processes like mTOR signaling and hormone secretion pathways. Our findings underscore the significant impact of RNA splicing on PanNET heterogeneity and suggest that detailed splicing profiles could serve as valuable tools for identifying novel biomarkers and therapeutic targets. This study provides crucial insights into PanNET molecular biology and paves the way for personalized therapies based on splicing features.

cancer biology↗

bioRxiv2019

3D imaging of colorectal cancer organoids identifies responses to Tankyrase inhibitors

Aberrant activation of the Wnt signalling pathway is required for tumour initiation and survival in the majority of colorectal cancers. The development of inhibitors of Wnt signalling has been the focus of multiple drug discovery programs targeting colorectal cancer and other malignancies associated with aberrant pathway activation. However, progression of new clinical entities targeting the Wnt pathway has been slow. One challenge lies with the limited predictive power of 2D cancer cell lines because they fail to fully recapitulate intratumoural phenotypic heterogeneity. In particular, the relationship between 2D cancer cell biology and cancer stem cell function is poorly understood. By contrast, 3D tumour organoids provide a platform in which complex cell-cell interactions can be studied. However, complex 3D models provide a challenging platform for the quantitative analysis of drug responses of therapies that have differential effects on tumour cell subpopulations. Here, we generated tumour organoids from colorectal cancer patients and tested their responses to inhibitors of Tankyrase (TNKSi) which are known to modulate Wnt signalling. Using compounds with 3 orders of magnitude difference in cellular mechanistic potency together with image-based assays, we demonstrate that morphometric analyses can capture subtle alterations in organoid responses to Wnt inhibitors that are consistent with activity against a cancer stem cell subpopulation. Overall our study highlights the value of phenotypic readouts as a quantitative method to asses drug-induced effects in a relevant preclinical model.

cancer biology↗

bioRxiv2019

The retinoblastoma tumor suppressor limits ribosomal readthrough during oncogene induced senescence

The origin and evolution of cancer cells is considered to be mainly fueled by mutations affecting the DNA sequence. Although errors in translation could also expand the cellular proteome, their role in cancer biology remains poorly understood. Tumor suppressors called \"caretakers\" block cancer initiation and progression by preventing DNA mutations and/or stimulating DNA repair. If translational errors contribute to tumorigenesis, then caretakers genes will prevent such errors in normal cells in response to oncogenic stimuli. Here, we show that the retinoblastoma protein (RB) acts as caretaker tumor suppressor by preventing the readthrough of termination codons, a process that allows proteins to be synthetized with additional domains. In particular, we show that expression of oncogenic ras in normal human cells triggers a cellular senescence response characterized by a significant reduction of basal ribosomal readthrough. However, inactivation of the RB tumor suppressor pathway in these cells, using the viral oncoprotein E7 or the oncogenic kinase CDK4 increased readthrough. Conversely, activation of the RB pathway by the tumor suppressor PML, the ribosomal proteins RPS14/uS11 and RPL22/eL22 or the CDK4/6 inhibitor palbociclib reduced readthrough. We thus reveal a novel function for the RB pathway as a caretaker of translational errors with implications for tumor suppression and cancer treatment.

molecular biology↗

bioRxiv2025

Identifying Molecular Determinants and Therapeutic Targets in Luminal B Breast Cancer: A Systems Biology Approach

Luminal B breast cancer (LBBC) is on the rise worldwide, with both incidence and mortality rates steadily increasing. Early detection proves difficult due to its aggressive characteristics, most notably its heightened proliferation rate and the complex interplay of molecular biomarkers, particularly in more advanced stages. In the present study, we conducted an in silico analysis of LBBC cell lines using the Gene Expression Omnibus (GEO) microarray dataset, which includes 30 LBBC and 11 normal samples. Differentially expressed genes (DEGs) were identified using RStudio. A series of analyses, including cancer data interrogation via pan-cancer analysis, eXpression2Kinases (X2K), and the Cancer Dependency Map (DepMAP), was carried out to elucidate the underlying signaling pathways, transcription factors (TFs), and kinases, as well as to perform stemformatics analysis. Protein-protein interaction (PPI) networks were reconstructed using STRING and Cytoscape, enabling the identification of co-expressed and hub genes through the cytoHubba plug-in. Of note, FGF2, EGFR, ADIPOQ, LIPE, MET, IGF1, FGF1, EGF, FGF7, and PPARG were identified as the top 10 driver genes; RELA, PPARG, CTCF, EGR1, and NFE2L2 were detected as predominant TFs in LBBC; and CDK1, CDK2, MAPK3, CSNK2A1, and MAPK14 were identified as potential biomarkers through hub gene clustering. Further analysis indicated hsa-mir-221-3p and hsa-mir-29a-3p as key miRNAs targeting LBBC-related genes. Collectively, our findings highlighted LBBC-associated genes, TFs, miRNAs, and pathways as prospective biomarkers, providing insights into LBBC diagnosis and therapeutic approaches.

bioinformatics↗

bioRxiv2020

Modulation of proteasome activity by curcumin and didemethylcurcumin

Modulation of proteasome function by pharmacological interventions and molecular biology tools is an active area of research in cancer biology and neurodegenerative diseases. Curcumin (diferuloylmethane) is a naturally occurring polyphenol that affects multiple signaling pathways and known to modulate PKC activities. The therapeutic significance of curcumin is often considered to reside in its anti-inflammatory, antioxidant, anti-angiogenic, or anti-apoptotic properties. However, recent research suggests that the therapeutic efficacy of curcumin may be due to its activity as a potent inhibitor of the proteasome. In this study, we show that both curcumin and its synthetic polyphenolic derivative (didemethylcurcumin), CUIII modulated proteasome activity in a biphasic manner. Curcumin and CUIII increased proteasome activity at nanomolar concentrations but inhibited proteasome activity at micromolar concentrations. Curcumin was more effective than CUIII in relative proteasome activity increase at nanomolar concentrations. Also, curcumin was more effective than CUIII in relative proteasome activity inhibition at micromolar concentration. The docking study was conducted on the 20S proteasome catalytic subunit. Estimated Kd values for curcumin and didemethylcurcumin are 0.0054M and 1.3167M, respectively. These values correlate well with the results of the effectiveness of modulation by curcumin compare to CUIII. The small size of CUIII makes its dock to the narrow cavity of the active site, but the binding interaction is not strong compare to curcumin. This study suggests that curcumin and its didemethyl derivative can be used to modulate proteasome activity. This communication implicates the reason why curcumin and its didemethyl derivative can be used to two different disease mechanisms, neurodegeneration, and cancer.

neuroscience↗

bioRxiv2022

A metabolic labeling-based chemoproteomic platform unravels the physiological roles of choline metabolites

Choline is an essential nutrient for mammalian cells. Our understanding of the cellular functions of choline and its metabolites, independent of their roles as choline lipid metabolism intermediates, remains limited. In addition to fundamental cellular physiology, this knowledge has implications for cancer biology because elevated choline metabolite levels are a hallmark of cancer. Here, we establish the mammalian choline metabolite-interacting proteome by utilizing a photocrosslinkable choline probe. To design this probe, we performed metabolic labeling experiments with structurally diverse choline analogs that resulted in the serendipitous discovery of a choline lipid headgroup remodeling mechanism involving sequential dealkylation and methylation steps. We demonstrate that phosphocholine inhibits the binding of one of the proteins identified, the attractive anticancer target, p32, to its endogenous ligands and to the promising p32-targeting anticancer agent, Lyp-1. Our results reveal that choline metabolites play vital roles in cellular physiology by serving as modulators of protein function.

biochemistry↗

bioRxiv2017

Clingen Cancer Somatic Working Group: standardizing and democratizing access to cancer molecular diagnostic data to drive translational research

A growing number of academic and community clinics are conducting genomic testing to inform treatment decisions for cancer patients (1). In the last 3-5 years, there has been a rapid increase in clinical use of next generation sequencing (NGS) based cancer molecular diagnostic (MolDx) testing (2). The increasing availability and decreasing cost of tumor genomic profiling means that physicians can now make treatment decisions armed with patient-specific genetic information. Accumulating research in the cancer biology field indicates that there is significant potential to improve cancer patient outcomes by effectively leveraging this rich source of genomic data in treatment planning (3). To achieve truly personalized medicine in oncology, it is critical to catalog cancer sequence variants from MolDx testing for their clinical relevance along with treatment information and patient outcomes, and to do so in a way that supports large-scale data aggregation and new hypothesis generation. One critical challenge to encoding variant data is adopting a standard of annotation of those variants that are clinically actionable. Through the NIH-funded Clinical Genome Resource (ClinGen) (4), in collaboration with NLMs ClinVar database and >50 academic and industry based cancer research organizations, we developed the Minimal Variant Level Data (MVLD) framework to standardize reporting and interpretation of drug associated alterations (5). We are currently involved in collaborative efforts to align the MVLD framework with parallel, complementary sequence variants interpretation clinical guidelines from the Association of Molecular Pathologists (AMP) for clinical labs (6). In order to truly democratize access to MolDx data for care and research needs, these standards must be harmonized to support sharing of clinical cancer variants. Here we describe the processes and methods developed within the ClinGens Somatic WG in collaboration with over 60 cancer care and research organizations as well as CLIA-certified, CAP-accredited clinical testing labs to develop standards for cancer variant interpretation and sharing.

bioinformatics↗

bioRxiv2025

Integrative network modeling of colorectal cancer reveals diagnostic signatures and therapeutic targets

Emerging evidence suggests that the interplay between multiple signaling pathways and the immune microenvironment influences tumorigenesis in cancers such as colorectal cancer (CRC). To gain an in-depth understanding of CRC mechanisms and identify novel therapeutic targets, we constructed a molecular interaction map (MIM) integrating key signaling pathways from cancer cells and immune cells within the tumor microenvironment. This map comprises 218 molecules and 328 interactions, curated using PubMed references and official gene symbols. We dynamically simulated the MIM for individual pathways and combinations via stimulus-response and perturbation analyses, calibrating the model with two CRC datasets: GSE1323 (primary tumor-to-metastasis progression) and GSE8671 (normal mucosa-to-adenoma progression), which served as experimental conditions. Simulations revealed distinct disease signatures for GSE1323, including (1) simultaneous activation of TNF/TNFRSF1A,B and EGF/EGFR with inactivation of ERE/ESR, and (2) simultaneous activation of TNF/TNFRSF1A,B and TLR4 with inactivation of ERE/ESR. For GSE8671, the signature was simultaneous activation of TNF/TNFRSF1A,B and TLR4. In silico perturbation analysis identified potent anti-cancer effects from concurrent inhibition of MAPK3 and STAT3 (GSE1323), and ELK1/ATF2 and STAT3 or MAPK14 and STAT3 (GSE8671), significantly reducing epithelial-mesenchymal transition (EMT), proliferation, and inflammation while increasing apoptosis. Disease signatures and therapeutic targets were validated using patient data through Kaplan-Meier survival analysis and machine learning. This integrative model recapitulates cancer biology, predicts biomarkers and therapeutic targets, and is extensible to other immunogenic cancers.

systems biology↗

bioRxiv2020

In situ characterization of the 3D microanatomy of the pancreas and pancreatic cancer at single cell resolution

Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest forms of cancer. Accumulating evidence indicates the tumor microenvironment is highly associated with tumorigenesis through regulation of cellular physiology, signaling systems, and gene expression profiles of cancer cells. Yet the mechanisms by which the microenvironment evolves from normal pancreas architecture to precursor lesions and invasive cancer is poorly understood. Obtaining high-content and high-resolution information from a complex tumor microenvironment in large volumetric landscapes represents a key challenge in the field of cancer biology. To address this challenge, we established a novel method to reconstruct three-dimensional (3D) centimeter-scale tissues containing billions of cells from serially sectioned histological samples, utilizing deep learning approaches to recognize eight distinct tissue subtypes from hematoxylin and eosin stained sections at micrometer and single-cell resolution. Using samples from a range of normal, precancerous, and invasive pancreatic cancer tissue, we map in 3D modes of cancer invasion in the tumor microenvironment, and emphasize the need for further 3D quantification of biological systems.

cancer biology↗

bioRxiv2025

Monotherapy cancer drug-blind response prediction is limited to intraclass generalization

Monotherapy cancer drug response prediction (DRP) models predict the response of a cell line to a given drug. Analyzing these models performance includes assessing their ability to predict the response of cell lines to new drugs, i.e. drugs that are not in the training set. Drug-blind prediction displays greatly diminished performance or outright failure across a wide range of model architectures and different large pharmacogenomic datasets. Drug-blind failure is hypothesized to be caused by the relatively limited set of drugs present in these datasets. The time and cost associated with further cell line experiments is significant, and it is impossible to predict beforehand how much data would be enough to overcome drug-blind failure. We must first define how current data contributes to drug-blind failure before attempting to remedy drug-blind failure with further data collection. In this work, we quantify the extent to which drug-blind generalizability relies on mechanistic overlap of drugs between training and testing splits. We first identify that the majority of mixed set DRP model performance can be attributed to drug overfitting, likely inhibiting generalization and preventing accurate analysis. Then, by specifically probing the drug-blind ability of models, we reveal the sources of generalizable drug features are confined to shared mechanisms of action and related pathways. Furthermore, we observed that, for certain mechanisms, we can significantly improve performance by limiting the training of models to a single mechanism compared to training on all drugs simultaneously. We conclude that drug-blind performance is a poor benchmark for DRP as it does not describe model behavior, it describes dataset behavior. Our investigation displays that deep learning models trained on large, monotherapy cell line panels can more accurately describe mechanism of action of drugs rather than their advertised connection to broader cancer biology. Author summaryIn this paper, we characterize the feature space of cancer drug-blind prediction. To understand the efficacy of a novel cancer drug it has never seen before (drug-blind), a model must be able to accurately compare this drug to all drugs it saw during training. These relationships between cancer drugs, the feature space, must be described well enough that this is possible. We believe that these relationships are poorly defined because cancer DRP models always display reduced performance in a drug-blind context. For the first time, we quantified the limits of generalization in a drug-blind setting. We showed that drug-blind generalization describes mechanistic relationships among drugs during model training. We also outlined new criteria with which to judge the drug-blind ability of a model. Failure of drug-blind prediction is an oft overlooked shortcoming in cancer DRP with potentially damaging downstream implications. We hope to show drug-blind ability of these models in a new light to guide others towards more pertinent tasks in cancer deep learning.

bioinformatics↗

bioRxiv2025

MTBP allosterically activates Cdk8-CycC kinase activity

How the approximately 300 human protein kinases identify their dedicated substrates despite the significant overlap of their phosphorylation consensus sequences is relevant for nearly all cellular processes. We show here that the Cdk8/19-CycC kinase uses mutually exclusive targeting and activation factors to facilitate distinct cellular roles. The Med12 protein is known to target Cdk8/19-CycC to the mediator of transcription complex to control the transcription of specific gene sets upon respective stimuli. We describe that a second Cdk8/19-CycC targeting factor exists, the replication origin firing regulator MTBP that targets the kinase to Med12-independent cellular roles. Both Med12 and MTBP constitute allosteric activators of the enzymatic Cdk8/19-CycC kinase activity in vitro. We describe the structural basis of this activation that involves distinct mechanisms how Med12 and MTBP reposition the T-loop of the kinase independently of T-loop phosphorylation - the canonical mechanism of CDK kinase activation. Our results support the following model: the Cdk8/19-CycC dimer alone has low enzymatic activity, which may help avoid off-target phosphorylation. Med12, MTBP and potentially other as yet unidentified accessory factors, target the kinase to distinct molecular environments, at the same time activating kinase activity for efficient substrate phosphorylation. Our work establishes an unusual mechanism of CDK kinase control and change the current paradigm how Cdk8/19-CycC selects its substrates. Substrate selection of the kinase may be relevant for cancer biology and therapy because Cdk8 is a well-established colorectal cancer promoting factor.

biochemistry↗

bioRxiv2020

FTO-mediated cytoplasmic m6Am demethylation adjusts stem-like properties in colorectal cancer cell

Cancer stem cells (CSCs) are a small but critical cell population for cancer biology since they display inherent resistance to standard therapies and give rise to metastases. Despite accruing evidence establishing a link between deregulation of epitranscriptome-related players and tumorigenic process, the role of messenger RNA (mRNA) modifications dynamic in the regulation of CSC properties remains poorly understood. Here, we show that the cytoplasmic pool of fat mass and obesity-associated protein (FTO) impedes CSC abilities in colorectal cancer through its m6Am (N6,2-O-dimethyladenosine) demethylase activity. While m6Am is strategically located next to the m7G-mRNA cap, its biological function is not well understood and has not been addressed in cancer. Low FTO expression in patient-derived cell lines elevates m6Am level in mRNA which results in enhanced in vivo tumorigenicity and chemoresistance. Inhibition of the nuclear m6Am methyltransferase, PCIF1/CAPAM, partially reverses this phenotype. FTO-mediated regulation of m6Am marking constitutes a novel, reversible pathway controlling CSC abilities that does not involve transcriptome remodeling, but could fine-tune translation efficiency of selected m6Am marked transcripts. Altogether, our findings bring to light the first biological function of the m6Am modification and its potential adverse consequences for colorectal cancer management.

cancer biology↗

bioRxiv2026

Oncogenes have the most distinct codon biases in the genome and codon signatures that oppose tumor suppressor genes

Oncogenes and tumor-suppressor genes play opposing roles in cancer biology to promote and restrict growth, respectively. Codon usage patterns interface with tRNA modifications to control translation, leading to gene-specific codon signatures with regulatory potential. As such, codon-biased translational regulation has been identified as a driver of proliferation and drug resistance in multiple cancers. We used advanced codon analytics methods to characterize and compare codon usage bias in oncogenes and tumor suppressor genes (TSGs) from humans and mice at group and gene-specific levels. We demonstrate that human oncogenes exhibit a distinct and opposing codon usage pattern to TSGs. This phenomenon is also present in mice but with less distinct oncogene bias relative to humans. Further comparison to 447 gene ontology groups demonstrated that human oncogenes have the most distinct codon usage patterns in the genome, while also highlighting that codon bias can separate functionally related genes and pathways from other biological processes. Using gene-specific codon analytics, we determined that human oncogenes have two types of extreme codon bias: a large group (N = 43) over-using G/C ending (GC3) codons and a smaller group (N = 12) over-using A/U (AU3) ending codons. While GC3 bias has been linked to increased translation in general, the AU3 finding suggests that genetic, environmental, or stress-related signals could drive the translation of this small group of oncogenes. The less extreme bias observed in mouse oncogenes and tumor suppressors likely underscores species-specific differences in oncogenic translation programs. Together, our findings highlight codon usage bias as a potential determinant of oncogene expression, provide a framework for ontology-based codon analysis, and uncover on species-specific differences in oncogene translation and codon usage biases.

cancer biology↗

bioRxiv2023

Combining evolution and protein language modelsfor an interpretable cancer driver mutationprediction with D2Deep

The mutations driving cancer are being increasingly exposed through tumor-specific genomic data. However, differentiating between cancer-causing driver mutations and random passenger mutations remains challenging. State-of-the-art homology-based predictors contain built-in biases and are often ill-suited to the intricacies of cancer biology. Protein Language Models have successfully addressed various biological problems but have not yet been tested on the challenging task of cancer driver mutation prediction at large scale. Additionally, they often fail to offer result interpretation, hindering their effective use in clinical settings. The AI-based D2Deep method we introduce here addresses these challenges by combining two powerful elements: i) a non-specialized protein language model that captures the makeup of all protein sequences and ii) protein-specific evolutionary information that encompasses functional requirements for a particular protein. D2Deep relies exclusively on sequence information, outperforms state-of-the-art predictors and captures intricate epistatic changes throughout the protein caused by mutations. These epistatic changes correlate with known mutations in the clinical setting and can be used for the interpretation of results. The model is trained on a balanced, somatic training set and so effectively mitigates biases related to hotspot mutations compared to state-of-the-art techniques. The versatility of D2Deep is illustrated by its performance on non-cancer mutation prediction, where most variants still lack known consequences. D2Deep predictions and confidence scores are available via https://tumorscope.be/d2deep to help with clinical interpretation and mutation prioritization.

bioinformatics↗