Biology subjects

Xie, W.

Publications and source records attributed to Xie, W..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

PPARgene 2.0: leveraging large language models and multi-omics data for enhanced identification and prediction of PPAR target genes

Peroxisome proliferator-activated receptors (PPARs) are ligand-activated transcription factors of the nuclear receptor superfamily. Upon ligand binding, PPARs activate target gene transcription and regulate a variety of important physiological processes such as lipid metabolism, inflammation, wound healing and immune responses. PPARgene is a database that integrates literature-curated and computationally predicted PPAR target genes. It provides gene-level annotations including tissue specificity, species, and supporting PubMed IDs. Computational predictions are generated using a machine learning method that combines PPRE motif analysis with microarray expression data. Here, we introduce PPARgene 2.0, a 10-year update to the original PPARgene database. This update adds 35 newly reported PPAR target genes, 20 PPAR{beta}/{delta} target genes, and 72 PPAR{gamma} target genes, bringing the total number of curated target genes in the database to 337. To retrieve newly reported PPAR target genes from the literature, we used two language models to screen publications after 2016. Candidate papers were then manually reviewed, and verified target genes were added to the updated database. This update also improved the predictive method by expanding the volume of high-throughput gene expression data and incorporating PPAR-related ChIP-seq datasets alongside in silico PPRE analysis. Fivefold cross-validation demonstrated that the new predictive method outperforms the original one. The updated prediction tool is available as part of the PPARgene 2.0 platform. The database is openly accessible at https://www.ppargene.org.

bioinformatics↗

bioRxiv2025

Beyond Inheritance: De novo Fast Motion Computation in Primate Visual Cortex

Objects move through space and time, generating sequential visuotopic activations in all sighted animals leading to motion perception of velocity defined by direction and speed. Humans can effortlessly see motion with speeds ranging from 0.25 to 500{degrees}/s. However, direction-selective neurons in the primary visual cortex (V1)--from which all subsequent processing is presumed to derive-- only encode directionality at low speeds. To resolve this paradox, we recorded neuronal responses to moving dots, gratings, and movies across the LGN, V1, MT and MST of the macaque motion pathway. Regardless of cell type and motion stimuli, V1 neurons lost direction selectivity at [~]29{degrees}/s while MT and MST neurons maintained it up to [~]82{degrees}/s and [~]183{degrees}/s, respectively. A cascaded spatiotemporal integration model reveals that at each cortex direction-selective neurons can generate velocity selectivity de novo, by integrating sequential visuotopic activations from preceding areas, irrespective of speed and directionality. By computing velocity anew, the primate brain effectively uses the cortical hierarchy itself to shift gears to efficiently encode slow and fast motion. Thus, five visual areas from the retina into the brains processing hierarchy, external spatiotemporal information is being computed afresh, offering insights for motion processing in other species, modalities and machine vision.

neuroscience↗

bioRxiv2025

A hidden protamine PTM code in sperm generates heterogeneous chromatin states and finetunes reproductive fitness

Traditionally, the sperm genome is thought to be packaged by protamines into a uniformly compact and inert chromatin structure. Here, we challenge this long-standing view by demonstrating that protamine post-translational modifications (PTMs) present on distinct protamine molecules create discrete protamine-DNA chromatin states, ranging from weak to tightly associated chromatin configurations. Loss of these modifications alters protamine-DNA interactions in vitro and in vivo, compromising sperm chromatin integrity and impairing fertility. Therefore, these findings demonstrate that protamines do not merely serve as inert packaging proteins; rather protamine PTMs establish functional heterogeneity within sperm chromatin, creating compartment-like domains analogous to those in somatic cells. Thus, PTMs allow protamines to do more than simply compact the paternal genome--they likely encode a molecular blueprint that orchestrates the timely unpacking and reorganization of the paternal genome after fertilization.

molecular biology↗

bioRxiv2025

Prognosis prediction using autophagy gene expression in osteosarcoma

Defining the transition from localized to metastatic osteosarcoma (OS) before overt dissemination is fundamental for improving survival. However, effective early diagnostic tools remain scarce, largely due to limited exploitation of the pre-metastatic tumor microenvironments own record of molecular barcode exposures encoded in cell-intrinsic stress states. We analyzed autophagy-dependent transcriptional datasets from 139 individuals to develop Auto-RS, a computational classifier that can integrate age and sex to deliver individualized risk management. The prognosis-interpretable features of Auto-RS recapitulate established molecular trajectories of metastasis at the single-cell level, capturing tumor cells intrinsic shift from proliferative to invasive states and revealing cooperative programs among cancer-associated fibroblasts and immune cells. More significantly, Auto-RS can expose chemotherapy vulnerabilities of newer drugs, providing a framework to prioritize therapeutics without direct testing in children. This framework brings a potential inflection point, where metastasis prediction and therapeutic stratification may converge to improve OS outcomes.

cell biology↗

bioRxiv2025

Deciphering the Antigenic Evolution of Seasonal Influenza A Viruses with PREDAC-Transformer: From Antigenic Clustering to Key Site Identification

Seasonal influenza viruses undergo continuous antigenic drift due to mutations in the hemagglutinin (HA) protein, rendering vaccines ineffective and posing a significant global public health challenge. Existing computational models can predict antigenic relationships but generally lack interpretability, making it difficult to reveal the molecular determinants driving antigenic changes. To address this, we developed PREDAC-Transformer, an end-to-end deep learning framework that integrates sequence, physicochemical, and evolutionary features, and employs a self-attention mechanism to capture long-range dependencies relevant to antigenicity. More importantly, we introduce the integrated antigenicity score, which combines attention-based attribution with information-theoretic metrics to provide continuous quantification and ranking of the antigenic contributions of individual amino acid site. Our results demonstrate that PREDAC-Transformer not only significantly improves the accuracy of antigenic relationship prediction but also successfully recapitulates major historical antigenic cluster transitions. Using integrated antigenicity score, we systematically identified two classes of key sites: global key sites with sustained impact on antigenic evolution, and cluster-transition determining sites that drive cluster transitions. These sites include most canonical epitopes and reveal additional functional residues previously overlooked, which may influence immune escape via cooperative effects or glycosylation. Collectively, these findings advance our understanding of influenza antigenic evolution and provide novel insights for refining computational models. PREDAC-Transformer achieves high-precision prediction while attributing antigenic differences to individual residues, thereby linking viral genomic variation, antigenic change, and public health decision-making. This framework has the potential to reduce experimental burdens in influenza surveillance and assist in vaccine strain recommendation, thereby supporting global influenza control efforts.

bioinformatics↗

bioRxiv2025

Multi-Scale Kinetics Modeling and Advanced Assay for mRNA-Lipid Nanoparticle Potency Assessment

mRNA lipid nanoparticle (mRNA-LNP) technology has attracted global attention, especially in vaccine development, due to its superior delivery efficiency, molecular stability, and safety profile. However, evaluating mRNA-LNP potency--defined as the quantifiable biological response elicited by the product--remains costly and time-consuming when relying solely on in vitro experiments. Rapid and reliable potency assessment is hindered by limited mechanistic understanding of delivery processes and sparse experimental data. To address these challenges, we present a mechanism-informed, multi-scale kinetic modeling framework that quantitatively captures the coupled dynamics across particle-level, cellular, and macroscopic scales. This model incorporates variability in LNP-cell interactions and accounts for critical factors such as dosage, LNP and cell size distributions, and cell proliferation dynamics--all of which influence delivery efficiency and response variability. By integrating advanced multi-omics assays--such as single-molecule fluorescent in situ hybridization (smFISH), which enables single-cell resolution of mRNA and protein expression--our framework leverages heterogeneous, multi-scale data to support mechanistically grounded modeling of mRNA delivery and enable robust predictions of therapeutic potency. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=143 SRC="FIGDIR/small/679406v1_ufig1.gif" ALT="Figure 1"> View larger version (63K): org.highwire.dtl.DTLVardef@1f3ec2borg.highwire.dtl.DTLVardef@1163893org.highwire.dtl.DTLVardef@1dc530borg.highwire.dtl.DTLVardef@1d0264a_HPS_FORMAT_FIGEXP M_FIG C_FIG

systems biology↗

bioRxiv2025

Harnessing the CD2 axis to broaden and enhance the efficacy of CAR T cell therapies

Patients with T-cell lymphomas and leukemias have overall poor outcomes due to the lack of targeted and effective treatments, particularly in the relapsed and refractory settings. Development of chimeric antigen receptor (CAR) T-cells against T-cell neoplasms is limited by a lack of discriminating T-cell antigens that allow for effective anti-tumor responses while preventing CAR T-cell fratricide. We hypothesized that targeting CD2, a pan-T-cell antigen, using anti-CD2 CAR T-cells engineered without CD2 expression (CART2), would support CAR T-cell manufacturability and preclinical efficacy. Optimized CD2-knockout CART2, generated using CRISPR-Cas9, eradicated primary patient-derived CD2+ hematological neoplasms in vitro and in vivo, secreted effector cytokines, and exhibited adequate proliferative capacity. Nevertheless, CD2 has a key costimulatory function, and its deletion could lead to CAR T-cell dysfunction. Therefore, we tested the role of the CD2:CD58 axis in CAR T-cells, using the anti-CD19 CART models. We demonstrate that CD2 loss attenuates CART19 efficacy by reducing avidity for tumor antigen, co-stimulation, and ultimately in vivo activity. Analogously, we show that tumor CD58 loss reduces CART19 efficacy. To overcome this issue, we developed a novel PD-1:CD2 switch receptor that rescues intracellular CD2 signaling, particularly when PD-L1 is engaged, resulting in improved in vivo outcomes. Collectively, we studied the role of CD2 both as a target for CAR T cell therapy and as a critical costimulatory protein, whose signaling can be rescued using the PD-1:CD2 switch receptor. This receptor can be incorporated into CAR T-cells and provides an effective strategy to overcome CD2-signaling deficiencies.

cancer biology↗

bioRxiv2025

Hypoimmunogenic human motor neurons induced from iPSCs in vivo substantially ameliorate ALS disease in large animal models

Stem cell-based therapy holds great potential for substituting degenerated motor neurons (MNs) in amyotrophic lateral sclerosis (ALS). Missing protocols for advanced differentiation of transplanted cells into MNs, immune rejection, and the lack of suitable ALS models for preclinical trials have slowed the development of effective therapies. Here, we employed multiplex genetic-editing to generate a novel human pluripotent stem cell line containing doxycycline (Dox)-inducible MNs-specific transcription factors and comprehensively modified immunomodulatory genes. We transplanted these cells into the spinal cord of ALS large animal models (SOD1G93A pigs and TIA1P362L rabbits), which faithfully recapitulate pathologies and symptoms observed in ALS patients. The transplanted cells could efficiently differentiate into functional MNs upon Dox treatment in vivo, distribute throughout the spinal cord and motor cortex via extensive migration, survive long-term without the need for immunosuppression. Notably, these MNs integrated into host neural circuits, as evidenced by their long projection of peripheral axons to target muscle and reformation of neuromuscular junctions. As result, pathologies and motor deficits were substantially ameliorated in both animal models. One Sentence SummaryHypoimmunogenic human motor neurons induced from iPSCs in vivo reform neuromuscular junctions and ameliorate ALS disease in pig and rabbit models.

neuroscience↗

bioRxiv2025

SARS-CoV-2 NSP14 inhibitor exhibits potent antiviral activity and reverses NSP14-driven host modulation

The emergence of SARS-CoV-2 variants and drug-resistant mutants highlights the urgent need for novel antiviral therapeutics. SARS-CoV-2 NSP14, an N7-guanosine methyltransferase, plays a critical role in viral RNA capping, enabling viral replication and immune evasion. While NSP14 has emerged as a promising drug target, its role in host-virus crosstalk and the cellular consequences of NSP14 inhibition remain poorly understood. Here, we present the identification and characterization of C10, a highly potent and selective first-in-class non-nucleoside inhibitor of the NSP14 S-adenosylmethionine (SAM)-binding pocket. C10 demonstrates robust antiviral activity against SARS-CoV-2, including its variants, with EC50 values ranging from 64.03 to 301.9 nM, comparable to the FDA-approved drug remdesivir in our cell-based assays. C10 also exhibits broad-spectrum activity against other betacoronaviruses and inhibits SARS-CoV-2 at the replication stage. C10 suppresses viral translation and exhibits immunostimulatory effect. Additionally, C10 specifically reversed NSP14-mediated alterations in host transcriptome. The antiviral efficacy of C10 was further validated in a transgenic mouse model of SARS-CoV-2 infection. Our findings highlight C10 as a promising candidate for the development of effective treatments against SARS-CoV-2 and its emerging variants. This study also uncovers a novel mechanism of NSP14 in SARS-CoV-2 pathogenesis and its therapeutic potential, providing insights that may extend to other viral capping methyltransferases.

microbiology↗

bioRxiv2025

Hedyotis Diffusae Herba Mitigates Rotenone-Induced Neurotoxicity via Akt Pathway in SH-SY5Y Cells

Hedyotis Diffusae Herba (HDH), known in traditional Chinese medicine for its heat-clearing and detoxifying properties, has been widely used to manage inflammation-related conditions. However, its neuroprotective potential and mechanism in the context of Parkinsons disease (PD) remain unclear.Network pharmacology and molecular docking were applied to predict the interactions between HDH components and PD-associated targets. Molecular dynamics simulations assessed the binding stability of major compounds to core targets. Rotenone-induced SH-SY5Y cell injury served as an in vitro PD model to evaluate HDHs biological effects. Cell viability, reactive oxygen species (ROS), IL-6 secretion, and Akt phosphorylation were assessed by CCK-8 assay, ROS detection kit, ELISA, and western blot, respectively. MK2206, an Akt inhibitor, was used to validate pathway involvement.Seven bioactive compounds were identified in HDH, among which stigmasterol and quercetin exhibited strong binding affinities with AKT1, IL-6, and JUN. Simulations confirmed stable interactions. HDH significantly increased cell viability and decreased intracellular ROS and IL-6 levels in rotenone-treated SH-SY5Y cells. Akt phosphorylation was restored by HDH, and the effect was blocked by MK2206, indicating Akt pathway participation.HDH exerts neuroprotective effects against rotenone-induced toxicity via antioxidant and anti-inflammatory activities and Akt pathway activation. These findings provide mechanistic insight and pharmacological support for HDH as a potential candidate in complementary therapy for PD.

cell biology↗

bioRxiv2025

H3K4me1 directs H3K36me2 and H3K36me3 deposition in land plants

Monomethylation of histone H3 lysine 4 (H3K4me1) marks enhancers in mammals. However, the function of H3K4me1 in plants remains largely unclear. Here, we analyzed the genome-wide distribution of H3K4me1 in diverse species across evolution, revealing a distinctive H3K4me1 distribution pattern in land plants. To explore the function of H3K4me1 in plants, we identified an H3K4me1-specific reader protein, Early heading date 3 (Ehd3), and solved the structure of Ehd3 in complex with the H3K4me1 peptide, revealing a unique binding module differing from the previously reported PHD finger proteins. We further identified an Ehd3-binding protein, SET domain group 724 (SDG724), and the deletion of either Ehd3 or SDG724 caused similar defects in plant phenotype and changes in transcriptome and epigenome profiles. Both Ehd3 and SDG724 are enriched at chromatin regions marked by H3K4me1 but not H3K4me2 or H3K4me3. Ehd3 activates the H3K36 methyltransferase SDG724, and H3K36me2/me3 are colocalized with H3K4me1 in the genomes of land plants. Collectively, our results reveal that H3K4me1 directs the establishment of H3K36me2 and H3K36me3 in land plants.

molecular biology↗

bioRxiv2025

Bridging Hearts and Muscles: Genetic Insights into Cardiac Structure and Sarcopenia

BackgroundCurrently, the link between cardiovascular diseases and sarcopenia is increasingly garnering attention from researchers. However, studies exploring the association between cardiac structure and sarcopenia are sparse. This study aims to investigate the potential genetic links between cardiac structural features and sarcopenia. MethodsDuring the discovery phase, we employed Linkage Disequilibrium Score Regression (LDSC) and Mendelian Randomization (MR) to assess the genetic correlation and causality between traits related to cardiac structure and sarcopenia, with further validation of the results using validation set data. In the study, we established a scoring system to identify high-confidence trait pairs. For these pairs, we conducted sensitivity analyses to assess their heterogeneity and pleiotropy. Additionally, we undertook follow-up studies of these trait pairs, including using mediation analysis to evaluate the potential mediating effects of lipids, and enrichment analysis to explore possible shared biological pathways linking these characteristics. ResultsThe genetic correlation analyses during the discovery and validation phases identified 5 pairs (in forward analysis) and 16 pairs (in reverse analysis) of high-confidence trait pairs. As a key sarcopenia-related feature, appendicular lean mass (ALM) exhibited positive causal relationships with several cardiac structural features, including the volumes of the left and right atria. Mediation analysis suggested that certain lipids, such as phosphatidylcholine, might mediate these causal relationships. Notably, gene and pathway enrichment analyses revealed that genes associated with significant SNPs in high-confidence trait pairs were enriched in multiple key biological processes and pathways, such as tube morphogenesis, cardiac right ventricle morphogenesis, and muscle system processes in the forward analysis, as well as skeletal system development, heart development, and hemopoiesis in the reverse analysis. Cell-type enrichment analysis pointed to endothelial cells, smooth muscle cells, fibroblasts, and mesenchymal stem cells. ConclusionsThis study provides evidence for genetic relationships between cardiac structures and sarcopenia-related traits. These findings may enhance our understanding of the biological mechanisms underlying age-related diseases and provide scientific basis for developing future preventive and therapeutic strategies.

genetics↗

bioRxiv2025

A reference genome assembly of the alpine forage grass Elymus nutans

Elymus nutans Griseb. (Poaceae: Triticeae, 2n=6x=42) is a dominant perennial plant species in the Qinghai-Tibetan Plateau in China, which is an important forage resource in high-altitude and cold regions and is the most popular species used for high yield artificial grassland planting because of its rich nutritional value and good palatability for herbivorous ruminant animals. In this study, using advanced sequencing technology, we generated an allohexaploid reference genome for E. nutans, representing the three sets of chromosomes (subgenomes St, Y, and H). This is the first study to our knowledge to confirm the origin of the Y subgenome, which shares a close relationship with the V subgenome. We predict that E. nutans arose via two-step hybridization and that its speciation occurred [~]3.16 MYA. The whole genome in this study will update the Triticeae family genomics information and provide theoretical basis for the evolutionary of its related speicies.

plant biology↗

bioRxiv2025

Automated Discovery of Therapeutic Biomaterial for Renally Impaired Hyperuricemia Patients by Natural Language Processing and Machine Learning

The exponential growth of scientific publications presents opportunities for researchers to identify valuable knowledge, especially in the highly interdisciplinary field --- biomaterials, where exploiting possible connections between unmet clinical needs and materials properties from literatures is crucial. However, with traditional literature reading, it is extremely challenging to marry unmet clinical needs with existing materials reported for different applications or other purposes. Here, to provide a not-renally cleared therapeutics for renally impaired hyperuricemia patients, we designed a multi-tiered framework MatWISE that fuses state-of-the-art natural language processing, semantic relationship mapping, and machine learning to automate the complex process of material discovery from a sea of scientific literatures published until December of 2022, and successfully identified and optimized {delta}-MnO2 into an orally administered, nonabsorbable uric acid (UA) lowering biomaterial. {delta}-MnO2 had superior serum and urine UA-lowering effect in three hyperuricemia mouse models, by comparing with a standard of care drug. {delta}-MnO2 is highly promising to serve as a safe and effective UA-lowering drug for renally impaired hyperuricemia patients. We demonstrated a new research paradigm for biomaterials that combining state-of-the-art machine learning techniques and a handful of experiments to discover a translationally relevant material from the massive existing research, for an unmet clinical need.

biochemistry↗

bioRxiv2025

Microbes under climatic refugia: Self-stabilizing subcommunity rank dynamics in large-river deltaic estuaries

Microorganisms, akin to macroorganisms, can be sheltered by climatic refugia, overlooked for optimism about microbial extinction. By categorizing microbial communities into ecologically distinct subcommunities, we conducted a two-year survey of large-river deltaic estuaries (LDEs) which are regarded as refugial hotspots, and established rank hierarchies to monitor temporal changes. We found that, consistent with macroecology, LDEs harbor distinct microbial community structures, with the abundance distribution of subcommunities exhibiting resilience against environmental perturbations, persisting on an annual cycle. In particular, while the evenness of subcommunity rank abundance curves fluctuates frequently, shifts in rank order are less dynamic but, once initiated, have a more pronounced impact on subsequent rank dynamics. Our results suggest that the microbial subcommunity rank structure in LDEs is both persistent and robust, aligning with classic patterns of species abundance distributions. This study emphasizes that, despite differences in their roles, the stable functioning of microbial rank systems is also fundamental to the refugial capacity of ecosystems and should be given equal importance.

ecology↗

bioRxiv2025

Mapping spatially-resolved transcriptomes in systemic sclerosis

Systemic sclerosis (SSc) is a prototypical fibrotic disease with high mortality and limited treatment options. Despite advances in single-cell RNA sequencing (scRNA-seq), the comprehensive understanding of cellular heterogeneity and cell-cell interaction within the fibrogenesis microenvironment remains limited. We generated spatially resolved transcriptome maps from healthy and SSc skin and built a scRNA-seq atlas to map the single-cell data to spatial space. This enabled us to identify a fibrotic niche, enriched with fibroblasts and macrophages, which is significantly expanded in SSc and correlated with clinical outcome. We revealed disease-specific cell states of fibroblasts and macrophages, and evaluated their spatial dependency on other cell types. We identified selective expression of ACKR3 in fibroblast progenitors that diminishes with SSc progression, which may serve to regulate CXCL12/CXCR4-mediated macrophage recruitment and fibrotic remodeling. Together, we provided an in-depth description at cellular and spatial levels of fine-tuned regulatory events occurring in SSc, offering spatiotemporal insights. One Sentence SummaryIntegrated spatial omics provide insight into the cellular and transcriptional landscape in spatially distinct microenvironments, which may drive fibrosis progression in SSc.

immunology↗

bioRxiv2025

Oral prodrug of a novel glutathione surrogate reverses metabolic dysregulation and attenuates neurodegenerative process in APP/PS1 mice

Glycation-induced oxidative stress underlies the numerous metabolic ravages of Alzheimers disease (AD). Reduced glutathione levels in AD lead to increased oxidative stress, including glycation-induced pathology. Previously, we showed that the accumulation of reactive 1,2-dicarbonyls such as methylglyoxal, the major precursor of non-enzymatic glycation products, was reduced by the increased function of GSH-dependent glyoxalase-1 enzyme in the brain. In this two-pronged study, we evaluate the therapeutic efficacy of an orally bioavailable prodrug of our lead glyoxalase substrate, pro-{psi}-GSH, for the first time in a transgenic Alzheimers disease mouse model. This prodrug delivers pharmacodynamically relevant brain concentrations of {psi}-GSH upon oral delivery. Chronic oral dosing of pro-{psi}-GSH effectively reverses the cognitive decline observed in the APP/PS1 mouse model. The prodrug successfully mirrors the robust effects of the parent drug i.e., reducing amyloid pathology, glycation stress, neuroinflammation, and the resultant neurodegeneration in these mice. We also report the first metabolomics study of such a treatment, which yields key biomarkers linked to the reversal of AD-related metabolic dysregulation. Collectively, this study establishes pro-{psi}-GSH as a viable, disease-modifying therapy for AD and paves the way for further preclinical advancement of such therapeutics. Metabolomic signatures identified could prove beneficial in the development of treatment-specific clinically translatable biomarkers. ABSTRACT GRAPHIC O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=196 SRC="FIGDIR/small/633247v1_ufig1.gif" ALT="Figure 1"> View larger version (60K): org.highwire.dtl.DTLVardef@a9c2c0org.highwire.dtl.DTLVardef@130ed71org.highwire.dtl.DTLVardef@134a19corg.highwire.dtl.DTLVardef@5e0f89_HPS_FORMAT_FIGEXP M_FIG C_FIG

neuroscience↗

bioRxiv2024

Drug Response Modeling across Cancers: Proteomics vs. Transcriptomics

Cancer cell lines are the most common in-vitro models for the evaluation of anti-cancer drug sensitivities. Past studies have been conducted to decipher and characterize the pharmacogenomic feature of cell lines based on other omics data, such as genomic mutation data and whole-genome RNA sequencing (RNA-seq) profiles. In particular, proteomic data is also an essential component for the characterization of tumours. However, different from RNA-seq datasets rich in numerous transcriptome profiles of cancer cell lines and cell viability assay of drug responses, the pharmacogenomic protein quantifications are relatively scarce. With the availability of the recently enriched proteomic dataset ProCan-DepMapSanger, we systematically evaluated the interplays among genomic mutations, transcription, and protein expressions across cancer cell lines. In general, blood cancers have higher RNA-protein correlations than those in solid cancers. The differential expression analysis on protein data helped identify more expressional and functional impact of genomic mutations of cancer genes. We also integrated the proteomic map with drug molecular chemical features to construct a bi-modal machine learning model to infer the drug sensitivities of cancer cell lines. Our results demonstrated that protein quantifications can lead to better drug response prediction performance than the model trained on transcriptome profiles. In addition, integrating protein data with drug chemical features, represented as molecular graphs and learned by Graph Neural Network, outperformed the state-of-the-art model DeepOmicNet for drug response prediction in proteomics.

bioinformatics↗