The rapid integration of genomic technologies in clinical diagnostics has resulted in the detection of a multitude of missense variants whose clinical significance is often unknown. As a result, a plethora of computational tools have been developed to facilitate variant interpretation. However, choosing an appropriate software from such a broad range of tools can be challenging; therefore, systematic benchmarking with high-quality, independent datasets is critical. Using three independent benchmarking datasets compiled from the ClinVar database, we evaluated the performance of ten widely used prediction algorithms with missense variants from 21 clinically relevant genes, including BRCA1 and BRCA2. A fourth dataset consisting of 1053 missense variants was also used to investigate the impact of type 1 circularity on their performance. The performance of the prediction algorithms varied widely across datasets. Based on Matthews Correlation Coefficient and Area Under the Curve, SNPs&GO and PMut consistently displayed an overall above-average performance across the datasets. Most of the tools demonstrated greater sensitivity and negative predictive values at the expense of lower specificity and positive predictive values. We also demonstrated that type 1 circularity significantly impacts the performance of these tools and, if not accounted for, may confound the selection of the best performing algorithms.

DNA damage tolerance (DDT) pathways mitigate the effects of DNA damage during replication by rescuing the replication fork stalled at a DNA lesion or other barriers and also repair discontinuities left in the newly replicated DNA. From yeast to mammalian cells, RAD18-regulated translesion synthesis (TLS) and template switching (TS) represent the dominant pathways of DDT. Monoubiquitylation of the polymerase sliding clamp PCNA by HRAD6A-B/RAD18, an E2/E3 protein pair, enables the recruitment of specialized TLS polymerases that can insert nucleotides opposite damaged template bases. Alternatively, the subsequent polyubiquitylation of monoubiquitin-PCNA by Ubc13-Mms2 (E2) and HLTF or SHPRH (E3) can lead to the switching of the synthesis from the damaged template to the undamaged newly synthesized sister strand to facilitate synthesis past the lesion. When immediate TLS or TS cannot occur, gaps may remain in the newly synthesized strand, partly due to the repriming activity of the PRIMPOL primase, which can be filled during the later phases of the cell cycle. The first part of this review will summarize the current knowledge about RAD18-dependent DDT pathways, while the second part will offer a molecular toolkit for the identification and characterization of the cellular functions of a DDT protein. In particular, we will focus on advanced techniques that can reveal single-stranded and double-stranded DNA gaps and their repair at the single-cell level as well as monitor the progression of single replication forks, such as the specific versions of the DNA fiber and comet assays. This collection of methods may serve as a powerful molecular toolkit to monitor the metabolism of gaps, detect the contribution of relevant pathways and molecular players, as well as characterize the effectiveness of potential inhibitors.

Background: Nimodipine, an L-type voltage-gated calcium channel blocker, achieves vasorelaxation by suppressing Ca ²⁺ -dependent activation of cerebrovascular smooth muscle cells and is used to prevent delayed ischemic deficit following subarachnoid hemorrhage. Our preclinical drug repurposing studies raised the possibility that nimodipine may attenuate the pro-inflammatory shift in microglial function in response to brain injury. We analyzed the effects of nimodipine on activated microglia at the level of morphological and functional phenotypes, as well as their transcriptomic profile. Methods: Live brain slice preparations from C57BL/6 mice and primary microglia cultures from the cortex of neonatal Sprague Dawley rats were used. Brain slices were subjected to ischemia, and microglial cultures were activated with lipopolysaccharide (LPS; 20 ng/ml). Both preparations were treated with nimodipine (5-10-20 μM). The degree of arborization was evaluated in Iba1-stained microglia and expressed as a transformation index (TI). Phagocytic activity of cultured microglia was visualized using fluorescent microbeads. TNFα levels in the cultures were measured with ELISA. Total RNA was isolated from microglia and processed for next generation RNA sequencing to determine differentially expressed genes. Results: Nimodipine suppressed the ameboid morphological transformation and increased phagocytosis triggered by ischemia in brain slices and LPS in microglia cultures. At the transcriptional level, LPS resulted in a pro-inflammatory microglial phenotype, affecting the expression of cytokines, the complement system and phagocytosis-related genes. Focusing on the role of calcium in microglial activation, LPS increased RNA transcription of ionotropic purinergic and some TRP channels but decreased the expression of voltage- and ligand-gated calcium channels. In the endoplasmic reticulum, LPS downregulated gene expression of Ryr and IP3 receptors and increased transcription of the SERCA calcium pump gene. Nimodipine co-administered with LPS altered the expression of 110 genes in the opposite direction to LPS activation, of which at least 20 were associated with microglial immune response, 7 with cell adhesion and 2 with autophagy regulation. Conclusion: The effect of nimodipine goes beyond cerebral vasorelaxation. Nimodipine attenuates microglial activation by modulating Ca ²⁺ -dependent gene expression involved in intracellular signaling cascades to drive microglial immune responses. Consideration should be given to expanding the medical field of indication of nimodipine.

Background: Circular RNAs (circRNAs) emerged as promising non-invasive cancer biomarkers due to their stability, abundance in body fluids, and regulatory potential. However, circRNA differential expression analysis (DEA) remains challenging, largely owing to lack of consensus on important preprocessing strategies such as filtering and normalization. While well-established bulk RNA-sequencing frameworks are commonly applied to circRNA data, newer approaches such as CIRI-DE (part of CIRI3 suite) integrate both linear and circular transcript information to improve detection. Despite developments, an assessment of these integrative strategies is lacking, and the critical impact of filtering on DEA model performance has not been comprehensively evaluated. Results: In this study, we evaluated the impact of multiple normalization and filtering strategies on circRNA DEA using five experimental datasets, including two in-house blood platelet sets and semi-parametric simulated in silico datasets. Our results emphasize the importance of selecting an appropriate filtering threshold, as overly lenient filtering substantially reduced model performance across datasets. We found edgeR’s filterByExpr() strategy particularly effective in handling zero counts in circRNA data, while also generating the most reliable results across most datasets. Furthermore, by incorporating linear and circular information as described in CIRI-DE, most methods identified a higher number of differentially expressed (DE) circRNAs compared to circular counts alone. Notably, circRNAs identified by both CIRI-DE and the modified bulk RNA-sequencing pipelines showed substantial overlap. Conclusion: Our findings demonstrate that automated filtering combined with linear-aware normalization significantly enhances the sensitivity and reproducibility of circRNA DEA, providing a standardized framework for more reliable biomarker discovery in transcriptomic research.

Genetic variants, particularly missense variants, play a significant role in human disease, contributing to the development of both monogenic disorders and cancer. Typically, these mutations can impair protein function by disrupting protein stability. Therefore, given their immense impact on human health, their identification and classification are a priority in clinical diagnostics and personalized medicine. Recent advancements in next-generation sequencing have significantly reduced sequencing costs. This has accelerated its integration into routine clinical diagnostics, resulting in the identification of thousands of missense variants, many of which have unknown impacts on protein function. These variants, known as variants of uncertain significance (VUS), present a major diagnostic challenge for medical professionals and patients. To address the increasing number of reported VUS, numerous in-silico methods have been developed over the past decade. However, their clinical application remains limited as they are currently accepted only as supportive evidence. Given their potential to swiftly prioritize and classify thousands of variants, continuous research is conducted to improve their predictive performance. Structural information has long been considered a valuable resource that could enhance the performance of these predictive models, but its integration has been largely hindered by the limited availability of protein structures. The development of AlphaFold2 has made it possible to access the structures of thousands of proteins, creating new opportunities to incorporate structural information into variant classification. In this thesis, we present a comprehensive overview of the performance of some of the most widely used variant effect predictors and explore the efficiency of structural features derived from mutated protein structures in improving missense variant classification. The first part of the study concentrated on the evaluation of the performance of ten variant effect predictors, PROVEAN, META-SNP, SIFT, PolyPhen-2 (HumDiv and HumVar), SNPs&GO, PredictSNP, PhD-SNP, PANTHER-PSEP, and PMut using general and gene-specific datasets. We have demonstrated that the performance of the benchmarked variant effect predictors varies considerably across different datasets, with some exhibiting gene-specific behavior. When analyzed within the framework of guidelines for missense variant classification we show that this performance influences the outcome of the computational analysis. Based on a set of criteria, we have determined the best-performing variant effect predictors for BRCA1 and BRCA2, which we recommend for variant prioritization and classification in these two genes. Additionally, we have highlighted the impact of type 1 circularity in the selection of the best-performing variant effect predictors, noting that failure to account for it can alter their ranking. In the second part of this study, we developed a large-scale protein structure prediction pipeline, along with a high-performance computing-optimized job submission strategy, to predict the structures of 77,713 proteins, including 65,612 variant models and 12,101 wild-type structures. With 70.1% of the generated structures predicted with high confidence, we present the largest collection of mutated protein structures to date. These structures may serve as a foundation for future studies in various areas in variant classification such as feature development and engineering as well as in structural bioinformatics, including studies on protein-protein interactions. Lastly, in part 3 we shifted our focus on exploring the potential of the generated mutated protein structures to enhance missense variant classification. By capturing differences between the wild-type and mutated, pathogenic, and benign structures five distinct features were developed: Alpha carbon distance (Cα-Dist), Alpha carbon delta pLDDT score (Cα-ΔpLDDT), Delta SASA Normalized (ΔSASA Normalized), Miyazawa-Jernigan Potential of the mutant (MJ-Mutant), and dRMS Local. These features were used to train three machine learning models, SIESTA, SIESTA-Str, and SIESTA-Seq. We have shown that while structure-derived features alone did not outperform sequence-based information, they have the potential to play a complementary role, as evidenced by the improved performance of SIESTA, which integrates both structural and sequence-based features. These findings along with the extended collection of mutated protein structures lay the groundwork for advancing future research in variant classification, with the potential to improve the classification and reclassification of variants of uncertain significance.