benchmarking

Fine-tuning genomic models on rare splicing events identifies a novel biomarker of TDP-43 pathology

Frustration Quenching and Network Topology of the Energy Landscape as Primary Determinants of Protein-Ligand Binding Pose Prediction by Deep Learning Models

This paper shows that deep learning models' performance in predicting protein-ligand binding poses depends on the energy landscape's frustration quenching and network topology, not algorithmic flaws.

User-friendly software for high-throughput processing of chlorophyll fluorescence images to assess plant stress responses: fvfmPy and fvfmR packages

The paper introduces fvfmPy and fvfmR, user-friendly tools for automated analysis of chlorophyll fluorescence images to assess plant stress responses, significantly reducing processing time compared to manual methods.

Comparative performance of reference-based metagenomic tools to identify species-level taxa among families of bacteria: benchmarking Mycobacteriaceae and Neisseriaceae

This study benchmarks reference-based metagenomic tools for species-level detection in closely related bacterial families, finding metapresence and sylph perform best.

Anti-target chemical-space coverage of Indian medicinal plant phytochemicals

The paper evaluates the chemical-space coverage of Indian medicinal plant phytochemicals against experimentally tested compounds for anti-target endpoints, finding most lack close structural representation.

Carbonic anhydrase inhibition rescues heart function in the Tg2576 Alzheimer's mouse model by ameliorating cardiac amyloid-β pathology and neurotrophic signaling impairment.

This study shows that carbonic anhydrase inhibitors improve heart function in Alzheimer's mice by reducing amyloid-β and restoring neurotrophic signaling.

Data-driven identification of the ecologically relevant climatic windows and climate summaries with StatClim model

The paper introduces StatClim, a Bayesian hierarchical model that identifies ecologically relevant climatic windows and summaries to better predict biological responses.

Quantification of cellular contributions to multi-cellular pattern formation

This paper introduces a geometric method to quantify individual cellular processes' contributions to multi-cellular pattern formation using synthetic and experimental data.

Lateral recurrence dynamically refines visual representations for task performance

This study shows that lateral recurrence in neural networks enhances identity separability in trained domains through dynamic refinement of representational geometry.

Comparative Analysis of Relative Ligand Binding Free Energy Simulation Methods: Amber-TI, GROMACS-NETI, OpenMM-FEP, and BLaDE-MSLD

This study compares four ligand binding free energy methods using the same force field and finds they show comparable accuracy with no significant performance differences.

Recording context and data splitting can inflate acoustic re-identification accuracy across species and studies

This paper shows that methodological choices can inflate acoustic re-identification accuracy across species and studies, and proposes diagnostics to assess true performance.

Multi-Omic and Spatial Profiling Defines Outcome-Associated Tumor States in Head and Neck Adenoid Cystic Carcinoma

The paper integrates multi-omic and spatial data from HNACC tumors to identify outcome-associated tumor states, finding that aggressive tumors adopt a MYC/E2F-driven luminal state and develop a four-gene classifier for survival prediction.

Artificial Intelligence Applied to Epigenomic Data in Leukemia: A Systematic Review of Diagnostic, Prognostic and Monitoring Models

This paper reviews AI models using epigenomic data for leukemia diagnosis, prognosis, and monitoring, finding high accuracy in classification but limited external validation and prospective testing.

Stemflow as a Storm-Driven Pathway for Fungal eDNA Transport

The paper uses ITS2 metabarcoding to analyze fungal eDNA in stemflow from three tree species, finding distinct communities influenced by host species and storm events.

Mechanistic simulation identifies predictive dose-dependent biomarkers of propofol anesthesia

The paper uses a mechanistic model to simulate propofol's effects on brain circuits, identifying a dose-dependent biomarker confirmed in empirical data.

Can carcasses cause environmental nucleic acid pollution? Implications for interpreting environmental RNA/DNA survey results

The paper investigates whether carcasses contribute to environmental RNA/DNA contamination and its implications for interpreting survey results.

Rapid simulation of organ-scale oxygen transport for tissue ischemia prediction

The paper introduces a computational toolkit to simulate organ-scale oxygen transport and predict tissue ischemia by analyzing vessel networks and physiological conditions.

Multidimensional Electroglottographic Characterization of Immediate Responses to Straw Phonation: A Pilot Methodological Study in Singers

This study examines the immediate electroglottographic effects of straw phonation in singers using four different straws, identifying potential biomarkers for further research.

Systematic mapping of enhancers, silencers, and topological elements with prime editing deletion screens

This paper introduces Swap-seq, a method to systematically delete and quantify the effects of enhancers, silencers, and topological elements on gene expression in human cells.

Large language models unlock the ecology of species interactions

This paper uses large language models to extract species interaction data from citizen science comments, expanding ecological data coverage and enabling new ecological research.

Illuminating dark genes and dark metabolites: calibrated functional association across kingdoms with autoencoders and correlation

VaLPAS pairs a bi-directional autoencoder with correlation for omics guilt-by-association with calibrated confidence; in L. starkeyi its autoencoder linked 176 unidentified metabolites to known-function proteins at ≥50% confidence.

Reactive and predictive processes during unpredictable driving collisions in virtual reality: an exploratory immersive study with multimodal neurophysiological monitoring

The paper investigates anticipatory neurophysiological responses to unpredictable collisions in VR using multimodal monitoring and finds a post-stimulus broadband power difference that correlates with collision outcomes.

Sketching the SRA: experiment-wide similarity search across 4.8 million metagenomes

The paper introduces resources for large-scale metagenome comparison in the SRA, revealing trends in microbial discovery and improving assembly quality.

Partner determination from protein sequences using class information

The paper introduces CLAPP, a method to determine paralog interactions by class, showing it can accurately assign interaction partners using weak or noisy scores when class labels are reliable.

DiverScan: attention-guided exploratory analysis of animal behaviour across diverse comparative experiments

DiverScan is a deep-learning framework that uses attention mechanisms to analyze animal behavior data, identifying condition-specific features across diverse experiments.

Improving the Resolution and Informativity of the Majority-Rule Consensus Tree Using Fine-Grained Dissimilarity Measures

The paper introduces methods to improve consensus tree resolution using fine-grained dissimilarity measures, showing better performance on datasets with low to moderate phylogenetic signal.

Perceptual decision-making: extending a standardized rodent task to humans

The paper compares perceptual decision-making strategies between humans and mice using a standardized task, finding that instructed humans perform better than uninstructed humans.

Evaluating Autonomous LLM Agents Across Molecular Prediction and Optimization Benchmarks

This paper evaluates autonomous LLM agents in molecular prediction and optimization benchmarks, showing they can develop methods that match or exceed human-developed approaches in several tasks.

Minimizer Density revisited: Models and Multiminimizers

This paper revisits the concept of density in k-mer-based sequence analysis, introduces multiminimizers as a novel technique, and proposes deduplicated density as a new metric.

Two Comparators May Be All We Need

Random forests on sequence and 2D chemistry pick the favored protein for a molecule, or the favored molecule for a protein, at 0.71 to 0.78 accuracy on unseen compounds.