Biology subjects

Florea, L.

Publications and source records attributed to Florea, L..

5 recordsLinked to original sources

bioRxiv2026

Benchmarking foundation models for splice site and exon annotation

Recent foundation and deep learning models have brought a generational leap in improving the quality of genome annotation, particularly in identifying genes and their structural elements, including exons and splice sites. However, they are trained on reduced datasets that may not capture biological complexity, such as differences between coding versus non-coding, terminal versus internal, constitutive versus alternatively spliced, and transposable element (TE)-derived exons. We evaluate several foundation models for gene and splice site annotation, including the transformer-based SegmentNT, Enformer and Borzoi, coupled with a segmentation head for per-base resolution, and the CNN-based SpliceAI and AlphaGenome, along with a newly developed fine-tuned model, STEP2h, on different classes of gene elements as described above. We found that the performance of all methods is highest for the class of exons found in their training data class and decreases drastically for classes of exons poorly represented. In particular, performance is highest for protein-coding genes, coding exons, and constitutive exons, and decreases drastically by up to 2-4 fold for non-coding internal exons, terminal exons, and exons that undergo alternative splicing. Similarly, performance is impaired on LINE-1 and Alu-derived exons. In contrast, a locally developed CNN model fine-tuned on a specialized TE-exon dataset showed improved performance in this category. Our study highlights the outstanding challenges in gene and exon annotation when leveraging powerful foundation models, and the need for further fine-tuning on judiciously selected classes of data or task-specific models to capture a broader, more diverse spectrum of gene features.

bioinformatics↗

bioRxiv2025

Chronic activation of a key exercise signal transducer, CaMKII, drives skeletal muscle aging and sarcopenia

Sarcopenia, the age-related loss of muscle strength and mass, contributes to adverse health outcomes in older adults. While exercise mitigates sarcopenia by transiently activating calcium (Ca2+)- and reactive oxygen species (ROS)-dependent signaling pathways that enhance muscle performance and adaptation, these same signals become chronically elevated in aged skeletal muscle and promote functional decline. Ca2+/calmodulin-dependent protein kinase II (CaMKII) is a key transducer of both Ca2+ and ROS signals during exercise. Here we show that CaMKII is chronically activated in aged muscles, promoting muscle dysfunction. Muscle-specific expression of a constitutively active CaMKII construct in young mice recapitulates features of aging muscles, including impaired contractility, progressive atrophy, mitochondrial disorganization, formation of tubular aggregates, and an older transcriptional profile characterized by the activation of inflammatory and stress response pathways. Mediation analysis identified altered heme metabolism as a potential mechanism of CaMKII-induced weakness, independent of muscle atrophy. Conversely, partial inhibition of CaMKII in aged muscle improved contractile function and shifted the transcriptome toward a more youthful state without inducing hypertrophy. These findings identify chronic CaMKII activation as a driver of functional and molecular muscle aging and support the concept that CaMKII exemplifies antagonistic pleiotropy, whereby its beneficial roles in promoting muscle performance and adaptation during youth may incur deleterious consequences in aging. We propose that persistent CaMKII activation in aged skeletal muscle reflects unresolved cellular stress and promotes maladaptive remodeling. Enhancing physiological reserve capacity through exercise, in combination with temporally targeted CaMKII inhibition, may help restore adaptive CaMKII signaling dynamics and preserve muscle function in aging.

physiology↗

bioRxiv2025

5-ethynyluridine perturbs nuclear RNA metabolism to promote the nuclear accumulation of TDP-43 and other RNA binding proteins

TDP-43, an essential nucleic acid binding protein and splicing regulator, is broadly disrupted in neurodegeneration. TDP-43 nuclear localization and function depend on the abundance of its nuclear RNA targets and its recruitment into large ribonucleoprotein complexes, which restricts TDP-43 nuclear efflux. To further investigate the interplay between TDP-43 and nascent RNAs, we aimed to employ 5-ethynyluridine (5EU), a widely used uridine analog for click chemistry labeling of newly transcribed RNAs. Surprisingly, 5EU induced the nuclear accumulation of TDP-43 and other RNA-binding proteins and attenuated TDP-43 mislocalization caused by disruption of the nuclear transport apparatus. RNA FISH demonstrated 5EU-induced nuclear accumulation of polyadenylated and GU-repeat-rich RNAs, suggesting increased retention of both processed and intronic RNAs. TDP-43 eCLIP confirmed that 5EU preserved TDP-43 binding at predominantly GU-rich intronic sites. RNAseq revealed significant 5EU-induced changes in alternative splicing, accompanied by an overall reduction in splicing diversity, without any major changes in RNA stability or TDP-43 splicing regulatory function. These data suggest that 5EU may impede RNA splicing efficiency and subsequent nuclear RNA processing and export. Our findings have important implications for studies utilizing 5EU and offer unexpected confirmation that the accumulation of endogenous nuclear RNAs promotes TDP-43 nuclear localization.

cell biology↗

bioRxiv2024

MntJULiP and Jutils: Differential splicing analysis of RNA-seq data with covariates

Differences in alternative splicing patterns can reveal important markers of phenotypic differentiation, including biomarkers of disease. Emerging large and complex RNA-seq datasets from disease and population studies include multiple confounders such as sex, age, ethnicity and clinical attributes, which demand highly specialized data analysis tools. However, few methods are equipped to handle the new challenges. We describe an implementation of our programs MntJULiP and Jutils for differential splicing detection and visualization from RNA-seq data that takes into account covariates. MntJULiP detects intron-level differences in alternative splicing from RNA-seq data using a Bayesian mixture model. Jutils visualizes alternative splicing variation with heatmaps, PCA and sashimi plots, and Venn diagrams. Our tools are scalable and can process thousands of samples within hours. We applied our methods to the collection of GTEx brain RNA-seq samples to deconvolute the effects of sex and age at death on the splicing patterns. In particular, clustering of covariate adjusted data identifies a subgroup of individuals undergoing a distinct splicing program during aging. MntJULiP and Jutils are implemented in Python and are available from https://github.com/splicebox/.

bioinformatics↗

bioRxiv2016

Splice Expression Variation Analysis (SEVA) for Differential Gene Isoform Usage in Cancer

MotivationCurrent bioinformatics methods to detect changes in gene isoform usage in distinct phenotypes compare the relative expected isoform usage in phenotypes. These statistics model differences in isoform usage in normal tissues, which have stable regulation of gene splicing. Pathological conditions, such as cancer, can have broken regulation of splicing that increases the heterogeneity of the expression of splice variants. Inferring events with such differential heterogeneity in gene isoform usage requires new statistical approaches.\n\nResultsWe introduce Splice Expression Variability Analysis (SEVA) to model increased heterogeneity of splice variant usage between conditions (e.g., tumor and normal samples). SEVA uses a rank-based multivariate statistic that compares the variability of junction expression profiles within one condition to the variability within another. Simulated data show that SEVA is unique in modeling heterogeneity of gene isoform usage, and benchmark SEVAs performance against EBSeq, DiffSplice, and rMATS that model differential isoform usage instead of heterogeneity. We confirm the accuracy of SEVAin identifying known splice variants in head and neck cancer and perform cross-study validation of novel splice variants. A novel comparison of splice variant heterogeneity between subtypes of head and neck cancer demonstrated unanticipated similarity between the heterogeneity of gene isoform usage in HPV-positive and HPV-negative subtypes and anticipated increased heterogeneity among HPV-negative samples with mutations in genes that regulate the splice variant machinery.\n\nConclusionThese results show that SEVA accurately models differential heterogeneity of gene isoform usage from RNA-seq data.\n\nAvailabilitySEVA is implemented in the R/Bioconductor package GSReg.\n\nContactbahman@jhu.edu, favorov@sensi.org, ejfertig@jhmi.edu

genomics↗