Large language models (LLMs) are rapidly becoming indispensable across the life‑sciences spectrum, from literature mining through clinical decision support to experimental design. Yet, in single‑cell RNA‑sequencing (scRNA‑seq) analysis, most LLM‑enabled tools remain opaque: they output a single label per cluster without disclosing the chain‑of‑ thought that led to that decision. This opaqueness undermines reproducibility, complicates peer‑review, and ultimately slows the adoption of otherwise powerful methods. We developed GPTBioInsightor, an LLM‑powered assistant that not only annotates cell types, cell states, and pathway activities but also narrates how it arrived at each conclusion, step-by-step. Across benchmark datasets-including peripheral blood mononuclear cells (PBMC3K) and pancreatic ductal adenocarcinoma-GPTBioInsightor achieved at least parity with expert manual curation while delivering transparent reasoning, confidence scores, and literature‑based evidence. By closing the "interpretability gap," GPTBioInsightor equips wet‑lab biologists, computational scientists, and reviewers with an audit‑ready trail, thereby accelerating discovery and fostering trust in AI‑assisted bioinformatics. GPTBioInsightor is freely available on GitHub under a BSD-3-Clause license (https://github.com/huang-sh/GPTBioInsightor).

Long-read RNA-seq uncovers complex transcriptome diversity, opening new avenues for isoform-level expression analysis. Nevertheless, the functional diversity of individual isoforms is still poorly understood. We introduce isoespy, an analysis pipeline for integrating isoform structures, differential expression, and functional annotations from long-read RNA-seq data. The workflow integrates third-party open reading frame predictors, juxtaposes isoform expression levels with gene models, and visualizes positional and non-positional user-provided features. We applied isoespy to a transcriptome dataset of hepatocellular carcinoma, identifying differences in isoform usage and predicted protein function. isoespy facilitates the interpretation of transcriptomic complexity through integrated structural and functional visualization. Isoespy is freely available at https://github.com/kolikem/isoespy.

Advances in live-cell fluorescence microscopy have enabled us to visualize single molecules (such as mRNAs and nascent proteins) in real time with high spatiotemporal resolution. However, these experiments generate large datasets that require complex computational processing pipelines to derive meaningful and quantitative information, which is a technical barrier for many researchers. Here, we introduce MicroLive, an open-source Python-based application for quantifying live-cell microscopy images. MicroLive provides an interactive Graphical User Interface (GUI) to perform key tasks, including cell segmentation, photobleaching correction, single-particle detection/tracking, spot intensity quantification, inter-channel colocalization, and time-series correlation analysis. As a ground-truth testing dataset, we used synthetic live-cell imaging data generated with the rSNAPed toolkit, demonstrating accurate extraction of biologically relevant parameters. Microscopy images of U-2 OS cells expressing a gene construct smHA-KDM5B-BoxB-MS2 were used to demonstrate the use of this software. MicroLive is distributed under a GPLv3 license and available on GitHub https://github.com/ningzhaoAnschutz/microlive. It can be installed via pip: pip install microlive.

Fluorosequencing generates millions of single peptide reads, yet a principled route to quantitative protein abundances has been lacking. We present a probabilistic framework that adapts expectation-maximization (EM) to the fluorosequencing measurement process, using posterior peptide probabilities from existing classifiers to estimate relative protein abundances. The algorithm iteratively updates abundances to maximize the likelihood of observed reads. We first evaluate five-protein simulations with realistic labeling and system errors. A simple Python implementation processes one million reads in under ten seconds on a standard workstation and reduces the mean absolute error by over an order of magnitude relative to a uniform-abundance guess, indicating robust performance in small-scale settings. We also assess scalability with full human-proteome simulations (20 642 proteins). Ten million reads are processed in under four hours on an NVIDIA DGX with a single Tesla V100 GPU, confirming tractability at proteome scale. Under current fluorosequencing error rates, the method yields modest accuracy gains, but when error rates are reduced, estimation error drops markedly, indicating that chemistry improvements would translate directly into more accurate quantitative proteomics. Overall, EM-based inference provides a scalable, model-driven bridge from peptide-level classification to protein-level quantification in fluorosequencing. Furthermore, the framework can also serve as a refinement step within other inference methods. The code and data utilized to produce all the results of this paper is at https://github.com/JavierKipen/ProtInfGPU.

MotivationSpecies-specific orphan genes lack homologues outside of a given taxon and frequently underlie unique species traits. Orphans can result from sequence divergence beyond recognition, when homologous proteins diverge to an extent at which sequence similarity search algorithms can no longer identify them as homologues, but they can also evolve de novo from previously noncoding sequences, in which case homologous protein-coding genes truly do not exist.ResultsHere we propose that sequence divergent orphans might be recognizable from their patterns of non-statistically significant similarity hits which are typically discarded. To test this, we simulated diverged orphan protein sequences under varying parameters. Using reversed protein sequences as negative control, we trained machine learning classifiers on features extracted from similarity search output. We found that this approach works, but performance of the models depends on the simulation parameters, with ∼90% accuracy when the underlying simulated divergence was moderate and ∼70% when it is extreme. When applying our classifiers on a set of real orphans we found that ∼30% of them are predicted to be divergent and these are shorter and more disordered than the rest. Our work contributes to the effort of better understanding how genetic novelty arises.Availability and implementationThe models and data used can be found at https://github.com/emiliostassios/Classification-of-divergent-genes-using-ML

MotivationThe integration of genome-scale metabolic and regulatory networks has received significant interest in cancer systems biology. However, the identification of lethal genetic interventions in these integrated models remains challenging due to the combinatorial explosion of potential solutions. To address this, we developed the genetic Minimal Cut Set (gMCS) framework, which computes synthetic lethal interactions—minimal sets of gene knockouts that are lethal for cellular proliferation- in genome-scale metabolic networks with signed directed acyclic regulatory pathways. Here, we present a novel formulation to calculate genetic Minimal Intervention Sets, gMISs, which incorporate both gene knockouts and knock-ins.ResultsWith our gMIS approach, we assessed the landscape of lethal genetic interactions in human cells, capturing interventions beyond synthetic lethality, including synthetic dosage lethality and tumor suppressor gene complexes. We applied the concept of synthetic dosage lethality to predict essential genes in cancer and demonstrated a significant increase in sensitivity when compared to large-scale gene knockout screen data. We also analyzed tumor suppressors in cancer cell lines and identified lethal gene knock-in strategies. Finally, we demonstrate how gMISs can help uncover potential therapeutic targets, providing examples in hematological malignancies.Availability and implementationThe gMCSpy Python package now includes gMIS functionalities. Access: https://github.com/PlanesLab/gMCSpy.

MotivationKlebsiella pneumoniae is a highly virulent superbug with rising antibiotic resistance worldwide. While matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) has transformed microbial identification, its application to antimicrobial resistance prediction remains underexplored, particularly for large clinical cohorts. In this study, we developed machine-learning models with feature-level interpretability using MALDI-TOF MS data to rapidly predict resistance to ciprofloxacin (CIP), cefuroxime (CXM), and ceftriaxone (CRO) in K. pneumoniae.ResultsUsing more than 28 000 isolates from two hospitals, the best-performing models reached an independent test accuracy of 0.7858, with sensitivity of 0.7289 and specificity of 0.8127. Several resistance-associated m/z signals—including 3657, 4341, 4519, 4709, 5070, 5409, 5921, 5939, and 6516—were consistently enriched in resistant isolates, offering interpretable spectral markers linked to resistance. Performance remained stable in time-based validation but declined across hospitals, suggesting sensitivity to geographic variability in resistance profiles. Overall, this study demonstrates that combining MALDI-TOF MS with machine learning enables rapid and interpretable prediction of resistance to commonly used fluoroquinolone and cephalosporins in K. pneumoniae. These findings highlight the clinical potential of such models for supporting empiric therapy and emphasize the importance of incorporating local data or adaptive strategies to improve generalizability across healthcare settings.Availability and implementationData available on request from the authors.

MotivationAnalysis of genomic variation in microbial genomes is crucial for understanding how microbes adapt, interact with their hosts, and influence health and disease. In metagenomic studies, where genetic material from entire microbial communities is sequenced, thousands of single-nucleotide polymorphisms can be detected across species and samples. However, identifying which of these variations has biologically or functionally relevant impacts remains a significant challenge.ResultsTo address this, we present SNPraefentia, a Python-based toolkit for prioritizing microbial SNPs based on their predicted functional relevance. The tool integrates multiple biologically meaningful parameters, including sequencing depth, physicochemical impact of amino acid substitutions, and the structural and functional context of mutations within annotated protein domains. SNPraefentia extracts variation depth and amino acid changes, annotates protein domains using UniProt, and computes individual impact scores. These are then integrated into a composite prioritization score that reflects the potential biological importance of each variant. Overall, SNPraefentia provides researchers with a systematic and reproducible approach to filter and rank microbial variants for downstream functional analysis or experimental validation.Availability and implementationThe toolkit and test data are freely available at https://github.com/muneebdev7/SNPraefentia.

SummaryMutual Hazard Networks (MHNs) are statistical models for analyzing (genetic) cancer progression. Many cancers develop silently and are only noticeable when they have significantly progressed, creating an observational gap until diagnosis. MHNs bridge this gap by reconstructing the underlying dynamics of disease progression. We present mhn, a Python package for dynamic cancer progression analysis using MHNs. It trains an MHN model from tumor genotypes. mhn overcomes challenges of numerical efficiency in model training by making use of state space restriction, allowing training MHNs with >100 mutational events, 5 times more than was possible before. The package offers (i) reconstruction of the most likely evolutionary history of tumors, (ii) sampling of artificial tumor histories, and (iii) visualization of genomic interactions and likely progression trajectories. These features substantially extend earlier implementations, providing a fast and user-friendly framework for researchers and clinicians to study cancer dynamics.Availability and implementationmhn can be installed from PyPI using pip and is available under the MIT License on GitHub (https://github.com/spang-lab/LearnMHN). Installation instructions and package functionalities are detailed on GitHub and PyPI, with a comprehensive guide on Read the Docs (https://learnmhn.readthedocs.io/en/latest/index.html) and a Jupyter notebook on GitHub to help users explore the package.

MotivationError-corrected next-generation sequencing (ECS) methods are increasingly used to assess mutagenicity and other genetic toxicology endpoints. The lack of open and standardized bioinformatic workflows and tools poses challenges to data reproducibility, comparability, and consistency in interpretation for its application in genetic toxicity assessment.ResultsWe present MutSeqR, an open source R package to analyse ECS mutation data for genetic toxicology studies. MutSeqR offers practical variant filtering, comparative analysis of mutation frequency between experimental conditions, dose–response assessment via benchmark dose calculations, mutation spectrum analysis, and clonality analyses. We demonstrate MutSeqR’s application using published datasets on mice treated with benzo[a]pyrene or benzo[b]fluoranthene, analysed using Duplex Sequencing and SMM-seq, respectively. MutSeqR’s flexible functions enable reproducible analyses across ECS platforms, facilitating research and regulatory applications in mutagenicity testing.Availability and implementationMutSeqR is freely available under an open source license at https://github.com/EHSRB-BSRSE-Bioinformatics/MutSeqR. Implemented in R (version 3.4.0 or greater), it supports all major operating systems. Sequencing data for Project 1 has been deposited in the Sequence Read Archive under accession number PRJNA803048. Variant call files for Project 2 are available on Mendeley Data (doi: 10.17632/65dnysxym8.1).