PyTorchKR Papers

최신 논문

총 139,336편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2504.20710공식 코드 · Papers with Code

SBMLtoOdin and Menelmacar: Interactive visualisation of systems biology models for expert and non-expert audiences

Motivation: Computational models in biology can increase our understanding of biological systems, be used to answer research questions, and make predictions. Accessibility and reusability of computational models is limited and often restricted to experts in programming and mathematics. This is due to the need to implement entire models and solvers from the mathematical notation models are normally presented as. Implementation: Here, we present SBMLtoOdin, an R package that translates differential equation models in SBML format from the BioModels database into executable R code using the R package odin, allowing researchers to easily reuse models. We also present Menelmacar, a a web-based application that provides interactive visualisations of these models by solving their differential equations in the browser. This platform allows non-experts to simulate and investigate models using an easy-to-use web interface. Availability: SBMLtoOdin is published under open source Apache 2.0 licence at https://github.com/bacpop/SBMLtoOdin and can be installed as an R package. The code for the Menelmacar website is published under MIT License at https://github.com/bacpop/odinviewer, and the website can be found at https://biomodels.bacpop.org/.

arXiv:2504.20103공식 코드 · Papers with Code

Heterogeneous network drug-target interaction prediction model based on graph wavelet transform and multi-level contrastive learning

Drug-target interaction (DTI) prediction is a core task in drug development and precision medicine in the biomedical field. However, traditional machine learning methods generally have the black box problem, which makes it difficult to reveal the deep correlation between the model decision mechanism and the interaction pattern between biological molecules. This study proposes a heterogeneous network drug target interaction prediction framework, integrating graph neural network and multi scale signal processing technology to construct a model with both efficient prediction and multi level interpretability. Its technical breakthroughs are mainly reflected in the following three dimensions:Local global feature collaborative perception module. Based on heterogeneous graph convolutional neural network (HGCN), a multi order neighbor aggregation strategy is designed.Multi scale graph signal decomposition and biological interpretation module. A deep hierarchical node feature transform (GWT) architecture is proposed.Contrastive learning combining multi dimensional perspectives and hierarchical representations. By comparing the learning models, the node representations from the two perspectives of HGCN and GWT are aligned and fused, so that the model can integrate multi dimensional information and improve the prediction robustness. Experimental results show that our framework shows excellent prediction performance on all datasets. This study provides a complete solution for drug target discovery from black box prediction to mechanism decoding, and its methodology has important reference value for modeling complex biomolecular interaction systems.

arXiv:2504.20102공식 코드 · Papers with Code

HyboWaveNet: Hyperbolic Graph Neural Networks with Multi-Scale Wavelet Transform for Protein-Protein Interaction Prediction

Protein-protein interactions (PPIs) are fundamental for deciphering cellular functions,disease pathways,and drug discovery.Although existing neural networks and machine learning methods have achieved high accuracy in PPI prediction,their black-box nature leads to a lack of causal interpretation of the prediction results and difficulty in capturing hierarchical geometries and multi-scale dynamic interaction patterns among proteins.To address these challenges, we propose HyboWaveNet,a novel deep learning framework that collaborates with hyperbolic graphical neural networks (HGNNs) and multiscale graphical wavelet transform for robust PPI prediction. Mapping protein features to Lorentz space simulates hierarchical topological relationships among biomolecules via a hyperbolic distance metric,enabling node feature representations that better fit biological a priori.HyboWaveNet inherently simulates hierarchical and scale-free biological relationships, while the integration of wavelet transforms enables adaptive extraction of local and global interaction features across different resolutions. Our framework generates node feature representations via a graph neural network under the Lorenz model and generates pairs of positive samples under multiple different views for comparative learning, followed by further feature extraction via multi-scale graph wavelet transforms to predict potential PPIs. Experiments on public datasets show that HyboWaveNet improves over both existing state-of-the-art methods. We also demonstrate through ablation experimental studies that the multi-scale graph wavelet transform module improves the predictive performance and generalization ability of HyboWaveNet. This work links geometric deep learning and signal processing to advance PPI prediction, providing a principled approach for analyzing complex biological systems

arXiv:2504.19813공식 코드 · Papers with Code

OmicsQ: A User-Friendly Platform for Interactive Quantitative Omics Data Analysis

Motivation: High-throughput omics technologies generate complex datasets with thousands of features that are quantified across multiple experimental conditions, but often suffer from incomplete measurements, missing values and individually fluctuating variances. This requires sophisticated analytical methods for accurate, deep and insightful biological interpretations, capable of dealing with a large variety of data properties and different amounts of completeness. Software to handle such data complexity is rare and mostly relies on programming-based environments, limiting accessibility for researchers without computational expertise. Results: We present OmicsQ, an interactive, web-based platform designed to streamline quantitative omics data analysis. OmicsQ integrates established statistical processing tools with an intuitive, browser-based visualization interface. It provides robust batch correction, automated experimental design annotation, and missing-data handling without imputation, which ensures data integrity and avoids artifacts from a priori assumptions. OmicsQ seamlessly interacts with external applications for statistical testing, clustering, analysis of protein complex behavior, and pathway enrichment, offering a comprehensive and flexible workflow from data import to biological interpretation that is broadly applicable tov data from different domains. Availability and Implementation: OmicsQ is implemented in R and R Shiny and is available at https://computproteomics.bmb.sdu.dk/app_direct/OmicsQ. Source code and installation instructions can be found at https://github.com/computproteomics/OmicsQ

arXiv:2504.19710공식 코드 · Papers with Code

PhyloProfile v2 -- Exploring multi-layered phylogenetic profiles at scale

Phylogenetic profiles visualize the presence-absence pattern of genes across taxa and are essential for delineating the evolutionary fate of genes and gene families. Integrating phylogenetic profiles across many genes and taxa reveals patterns of coevolution, aiding the predictions of gene functions and interactions. The surge of genome sequences generated by biodiversity genomics projects allows to compile phylogenetic profiles at an unprecedented scale. PhyloProfile v2 was designed to cope with the novel challenges of visualizing and analyzing phylogenetic profiles comprising millions of pairwise orthology relationships. By providing the ability to interact with the visualization and dynamically filter the data, PhyloProfile v2 facilitates a seamless transition from survey analyses across thousands of genes and taxa down to the feature architecture comparison of two ortholog pairs within the same analysis. As one key innovation, PhyloProfile v2 allows the display of phylogenetic profiles in 2D or 3D using dimensionality reduction techniques. This novel perspective eases, for example, the identification of taxa with similar presence/absence patterns of genes irrespective of their phylogenetic relationships. PhyloProfile v2 is available as an R package at Bioconductor https://doi.org/doi:10.18129/B9.bioc.PhyloProfile. The open-source code and documentation are provided under MIT license at https://github.com/BIONF/PhyloProfile

arXiv:2504.18076공식 코드 · Papers with Code

3plex Web: An Interactive Platform for RNA:DNA Triplex Prediction and Analysis

Summary: Long non-coding RNAs (lncRNAs) exert their functions by cooperating with other molecules including proteins and DNA. Triplexes, formed through the interaction between a single-stranded RNA (ssRNA) and a double-stranded DNA (dsDNA), have been consistently described as a mechanism that allows lncRNAs to target specific genomic sequences in vivo. Building on the computational tool 3plex, we developed 3plex Web, an accessible platform that enhances RNA:DNA triplex prediction by integrating interactive visualization, statistical evaluation, and user-friendly downstream analysis workflows. 3plex Web implements new features such as input randomization for statistical assessments, interactive profile plotting for triplex stability, and customizable DNA Binding Domain (DBD) selection. This platform enables rapid analysis through PATO, substantially reducing processing times compared to previous methods, while offering Snakemake workflows to integrate gene expression data and explore lncRNA regulatory mechanisms. Availability and implementation: 3plex Web is freely available at https://3plex.unito.it as an online web service. The source code for 3plex is available at https://github.com/molinerisLab/3plex, paired with a definition file to set up the application into a Singularity image. Contact: [email omitted] Keywords: DNA; RNA; RNA-DNA interaction; triplex; long non-coding RNA; lncRNA; gene regulation; web application

arXiv:2504.17568공식 코드 · Papers with Code

Beyond Cox Models: Assessing the Performance of Machine-Learning Methods in Non-Proportional Hazards and Non-Linear Survival Analysis

Survival analysis often relies on Cox models, assuming both linearity and proportional hazards (PH). This study evaluates machine and deep learning methods that relax these constraints, comparing their performance with penalized Cox models on a benchmark of three synthetic and three real datasets. In total, eight different models were tested, including six non-linear models of which four were also non-PH. Although Cox regression often yielded satisfactory performance, we showed the conditions under which machine and deep learning models can perform better. Indeed, the performance of these methods has often been underestimated due to the improper use of Harrell's concordance index (C-index) instead of more appropriate scores such as Antolini's concordance index, which generalizes C-index in cases where the PH assumption does not hold. In addition, since occasionally high C-index models happen to be badly calibrated, combining Antolini's C-index with Brier's score is useful to assess the overall performance of a survival method. Results on our benchmark data showed that survival prediction should be approached by testing different methods to select the most appropriate one according to sample size, non-linearity and non-PH conditions. To allow an easy reproducibility of these tests on our benchmark data, code and documentation are freely available at https://github.com/compbiomed-unito/survhive.

arXiv:2504.16096공식 코드 · Papers with Code

BrainPrompt: Multi-Level Brain Prompt Enhancement for Neurological Condition Identification

Neurological conditions, such as Alzheimer's Disease, are challenging to diagnose, particularly in the early stages where symptoms closely resemble healthy controls. Existing brain network analysis methods primarily focus on graph-based models that rely solely on imaging data, which may overlook important non-imaging factors and limit the model's predictive power and interpretability. In this paper, we present BrainPrompt, an innovative framework that enhances Graph Neural Networks (GNNs) by integrating Large Language Models (LLMs) with knowledge-driven prompts, enabling more effective capture of complex, non-imaging information and external knowledge for neurological disease identification. BrainPrompt integrates three types of knowledge-driven prompts: (1) ROI-level prompts to encode the identity and function of each brain region, (2) subject-level prompts that incorporate demographic information, and (3) disease-level prompts to capture the temporal progression of disease. By leveraging these multi-level prompts, BrainPrompt effectively harnesses knowledge-enhanced multi-modal information from LLMs, enhancing the model's capability to predict neurological disease stages and meanwhile offers more interpretable results. We evaluate BrainPrompt on two resting-state functional Magnetic Resonance Imaging (fMRI) datasets from neurological disorders, showing its superiority over state-of-the-art methods. Additionally, a biomarker study demonstrates the framework's ability to extract valuable and interpretable information aligned with domain knowledge in neuroscience. The code is available at https://github.com/AngusMonroe/BrainPrompt

arXiv:2504.15855공식 코드 · Papers with Code

Generating heterogeneous data on gene trees

We introduce GenPhylo, a Python module that simulates nucleotide sequence data along a phylogeny avoiding the restriction of continuous-time Markov processes. GenPhylo uses directly a general Markov model and therefore naturally incorporates heterogeneity across lineages. We solve the challenge of generating transition matrices with a pre-given expected number of substitutions (the branch length information) by providing an algorithm that can be incorporated in other simulation software.

arXiv:2504.14345공식 코드 · Papers with Code

Integrating LLM-Generated Views into Mean-Variance Optimization Using the Black-Litterman Model

Portfolio optimization faces challenges due to the sensitivity in traditional mean-variance models. The Black-Litterman model mitigates this by integrating investor views, but defining these views remains difficult. This study explores the integration of large language models (LLMs) generated views into portfolio optimization using the Black-Litterman framework. Our method leverages LLMs to estimate expected stock returns from historical prices and company metadata, incorporating uncertainty through the variance in predictions. We conduct a backtest of the LLM-optimized portfolios from June 2024 to February 2025, rebalancing biweekly using the previous two weeks of price data. As baselines, we compare against the S&P 500, an equal-weighted portfolio, and a traditional mean-variance optimized portfolio constructed using the same set of stocks. Empirical results suggest that different LLMs exhibit varying levels of predictive optimism and confidence stability, which impact portfolio performance. The source code and data are available at https://github.com/youngandbin/LLM-MVO-BLM.

arXiv:2505.02717공식 코드 · Papers with Code

Topology across Scales on Heterogeneous Cell Data

Multiplexed imaging allows multiple cell types to be simultaneously visualised in a single tissue sample, generating unprecedented amounts of spatially-resolved, biological data. In topological data analysis, persistent homology provides multiscale descriptors of ``shape" suitable for the analysis of such spatial data. Here we propose a novel visualisation of persistence homology (PH) and fine-tune vectorisations thereof (exploring the effect of different weightings for persistence images, a prominent vectorisation of PH). These approaches offer new biological interpretations and promising avenues for improving the analysis of complex spatial biological data especially in multiple cell type data. To illustrate our methods, we apply them to a lung data set from fatal cases of COVID-19 and a data set from lupus murine spleen.

arXiv:2505.01348공식 코드 · Papers with Code

Learning Stabilizing Policies via an Unstable Subspace Representation

We study the problem of learning to stabilize (LTS) a linear time-invariant (LTI) system. Policy gradient (PG) methods for control assume access to an initial stabilizing policy. However, designing such a policy for an unknown system is one of the most fundamental problems in control, and it may be as hard as learning the optimal policy itself. Existing work on the LTS problem requires large data as it scales quadratically with the ambient dimension. We propose a two-phase approach that first learns the left unstable subspace of the system and then solves a series of discounted linear quadratic regulator (LQR) problems on the learned unstable subspace, targeting to stabilize only the system's unstable dynamics and reduce the effective dimension of the control space. We provide non-asymptotic guarantees for both phases and demonstrate that operating on the unstable subspace reduces sample complexity. In particular, when the number of unstable modes is much smaller than the state dimension, our analysis reveals that LTS on the unstable subspace substantially speeds up the stabilization process. Numerical experiments are provided to support this sample complexity reduction achieved by our approach.

arXiv:2505.01277공식 코드 · Papers with Code

Scoring-Assisted Generative Exploration for Proteins (SAGE-Prot): A Framework for Multi-Objective Protein Optimization via Iterative Sequence Generation and Evaluation

Proteins play essential roles in nature, from catalyzing biochemical reactions to binding specific targets. Advances in protein engineering have the potential to revolutionize biotechnology and healthcare by designing proteins with tailored properties. Machine learning and generative models have transformed protein design by enabling the exploration of vast sequence-function landscapes. Here, we introduce Scoring-Assisted Generative Exploration for Proteins (SAGE-Prot), a framework that iteratively combines autoregressive protein generation with quantitative structure-property relationship models for fine-tuned optimization. By integrating diverse protein descriptors, SAGE-Prot enhances key properties, including binding affinity, thermal stability, enzymatic activity, and solubility. We demonstrate its effectiveness by optimizing GB1 for binding affinity and thermal stability and TEM-1 for enzymatic activity and solubility. Leveraging curriculum learning, SAGE-Prot adapts rapidly to increasingly complex design objectives, building on past successes. Experimental validation demonstrated that SAGE-Prot-generated proteins substantially outperformed their wild-type counterparts, achieving up to a 17-fold increase in beta-lactamase activity, underscoring SAGE-Prot's potential to tackle critical challenges in protein engineering. As generative models continue to evolve, approaches like SAGE-Prot will be indispensable for advancing rational protein design.

arXiv:2505.01149공식 코드 · Papers with Code

Spatial modeling of forest-savanna bistability: Impacts of fire dynamics and timescale separation

Forest-savanna bistability - the hypothesis that forests and savannas exist as alternative stable states in the tropics - and its implications are key challenges for mathematical modelers and ecologists in the context of ongoing climate change. To generate new insights into this problem, we present a spatial Markov jump process model of savanna forest fires that integrates key ecological processes, including seed dispersal, fire spread, and non-linear vegetation flammability. In contrast to many models of forest-savanna bistability, we explicitly model both fire dynamics and vegetation regrowth in a mathematically tractable framework. This approach bridges the gap between slow-timescale vegetation models and highly resolved fire dynamics, shedding light on the influence of short-term and transient processes on vegetation cover. In our spatial stochastic model, bistability arises from periodic fires that maintain low forest cover, whereas dense forest areas inhibit fire spread and preserve high tree density. The deterministic mean-field approximation of the model similarly predicts bistability, but deviates quantitatively from the fully spatial model, especially in terms of its transient dynamics. These results also underscore the critical role of timescale separation between fire and vegetation processes in shaping ecosystem structure and resilience.

arXiv:2505.00719공식 코드 · Papers with Code

p2smi: A Python Toolkit for Peptide FASTA-to-SMILES Conversion and Molecular Property Analysis

Converting peptide sequences into useful representations for downstream analysis is a common step in computational modeling and cheminformatics. Furthermore, peptide drugs (e.g., Semaglutide, Degarelix) often take advantage of the diverse chemistries found in noncanonical amino acids (NCAAs), altered stereochemistry, and backbone modifications. Despite there being several chemoinformatics toolkits, none are tailored to the task of converting a modified peptide from an amino acid representation to the chemical string nomenclature Simplified Molecular-Input Line-Entry System (SMILES), often used in chemical modeling. Here we present p2smi, a Python toolkit with CLI, designed to facilitate the conversion of peptide sequences into chemical SMILES strings. By supporting both cyclic and linear peptides, including those with NCAAs, p2smi enables researchers to generate accurate SMILES strings for drug-like peptides, reducing the overhead for computational modeling and cheminformatics analyses. The toolkit also offers functionalities for chemical modification, synthesis feasibility evaluation, and calculation of molecular properties such as hydrophobicity, topological polar surface area, molecular weight, and adherence to Lipinski's rules for drug-likeness.

arXiv:2505.00650공식 코드 · Papers with Code

OmicsCL: Unsupervised Contrastive Learning for Cancer Subtype Discovery and Survival Stratification

Unsupervised learning of disease subtypes from multi-omics data presents a significant opportunity for advancing personalized medicine. We introduce OmicsCL, a modular contrastive learning framework that jointly embeds heterogeneous omics modalities-such as gene expression, DNA methylation, and miRNA expression-into a unified latent space. Our method incorporates a survival-aware contrastive loss that encourages the model to learn representations aligned with survival-related patterns, without relying on labeled outcomes. Evaluated on the TCGA BRCA dataset, OmicsCL uncovers clinically meaningful clusters and achieves strong unsupervised concordance with patient survival. The framework demonstrates robustness across hyperparameter configurations and can be tuned to prioritize either subtype coherence or survival stratification. Ablation studies confirm that integrating survival-aware loss significantly enhances the predictive power of learned embeddings. These results highlight the promise of contrastive objectives for biological insight discovery in high-dimensional, heterogeneous omics data.

arXiv:2504.20561공식 코드 · Papers with Code

BEER: Biochemical Estimator and Explorer of Residues -- A Comprehensive Software Suite for Protein Sequence Analysis

Protein sequence analysis underpins research in biophysics, computational biology, and bioinformatics. We introduce BEER, a crossplatform graphical interface that accepts FASTA or Protein Data Bank (PDB) files, or manual sequence entry, and instantly computes a suite of physicochemical metrics, such as amino acid composition, Kyte Doolittle hydrophobicity profiles, net charge versus pH curves with automatic isoelectric point determination, solubility predictions, and key indices such as molecular weight, extinction coefficient, GRAVY (grand average of hydropathicity) score, instability index, and aromaticity. BEER's interactive visualizations, including bar and pie charts, hydropathy plots, residue level bead models, and radar diagrams make it easy to explore physicochemical properties of protein chains. A multichain module also enables direct comparison of complex assemblies. Built in Python with BioPython, PyQt5, and matplotlib, BEER delivers complete analyses of sequences up to 10000 residues in under one second.

arXiv:2504.20127공식 코드 · Papers with Code

Learning Hierarchical Interaction for Accurate Molecular Property Prediction

Discovering molecules with desirable molecular properties, including ADMET profiles, is of great importance in drug discovery. Existing approaches typically employ deep learning models, such as Graph Neural Networks (GNNs) and Transformers, to predict these molecular properties by learning from diverse chemical information. However, these models often fail to efficiently capture and utilize the hierarchical nature of molecular structures, and often lack mechanisms for effective interaction among multi-level features. To address these limitations, we propose a Hierarchical Interaction Message Passing Mechanism, which serves as the foundation of our novel model, the Hierarchical Interaction Message Net (HimNet). Our method enables interaction-aware representation learning across atomic, motif, and molecular levels via hierarchical attention-guided message passing. This design allows HimNet to effectively balance global and local information, ensuring rich and task-relevant feature extraction for downstream property prediction tasks, such as Blood-Brain Barrier Permeability (BBBP). We systematically evaluate HimNet on eleven datasets, including eight widely-used MoleculeNet benchmarks and three challenging, high-value datasets for metabolic stability, malaria activity, and liver microsomal clearance, covering a broad range of pharmacologically relevant properties. Extensive experiments demonstrate that HimNet achieves the best or near-best performance in most molecular property prediction tasks. Furthermore, our method exhibits promising hierarchical interpretability, aligning well with chemical intuition on representative molecules. We believe that HimNet offers an accurate and efficient solution for molecular activity and ADMET property prediction, contributing to advanced decision-making in the early stages of drug discovery.

arXiv:2504.20096공식 코드 · Papers with Code

Towards Practical Second-Order Optimizers in Deep Learning: Insights from Fisher Information Analysis

First-order optimization methods remain the standard for training deep neural networks (DNNs). Optimizers like Adam incorporate limited curvature information by preconditioning the stochastic gradient with a diagonal matrix. Despite the widespread adoption of first-order methods, second-order optimization algorithms often exhibit superior convergence compared to methods like Adam and SGD. However, their practicality in training DNNs is still limited by a significantly higher per-iteration computational cost compared to first-order methods. In this thesis, we present AdaFisher, a novel adaptive second-order optimizer that leverages a diagonal block-Kronecker approximation of the Fisher information matrix to adaptively precondition gradients. AdaFisher aims to bridge the gap between the improved convergence and generalization of second-order methods and the computational efficiency needed for training DNNs. Despite the traditionally slower speed of second-order optimizers, AdaFisher is effective for tasks such as image classification and language modeling, exhibiting remarkable stability and robustness during hyperparameter tuning. We demonstrate that AdaFisher outperforms state-of-the-art optimizers in both accuracy and convergence speed. The code is available from https://github.com/AtlasAnalyticsLab/AdaFisher.

arXiv:2504.19933공식 코드 · Papers with Code

Automated decision-making for dynamic task assignment at scale

The Dynamic Task Assignment Problem (DTAP) concerns matching resources to tasks in real time while minimizing some objectives, like resource costs or task cycle time. In this work, we consider a DTAP variant where every task is a case composed of a stochastic sequence of activities. The DTAP, in this case, involves the decision of which employee to assign to which activity to process requests as quickly as possible. In recent years, Deep Reinforcement Learning (DRL) has emerged as a promising tool for tackling this DTAP variant, but most research is limited to solving small-scale, synthetic problems, neglecting the challenges posed by real-world use cases. To bridge this gap, this work proposes a DRL-based Decision Support System (DSS) for real-world scale DTAPS. To this end, we introduce a DRL agent with two novel elements: a graph structure for observations and actions that can effectively represent any DTAP and a reward function that is provably equivalent to the objective of minimizing the average cycle time of tasks. The combination of these two novelties allows the agent to learn effective and generalizable assignment policies for real-world scale DTAPs. The proposed DSS is evaluated on five DTAP instances whose parameters are extracted from real-world logs through process mining. The experimental evaluation shows how the proposed DRL agent matches or outperforms the best baseline in all DTAP instances and generalizes on different time horizons and across instances.