Biology subjects

Uvarovskii, A.

Publications and source records attributed to Uvarovskii, A..

4 recordsLinked to original sources

bioRxiv2025

Cutting through the clutter: minimizing redundancy in GO enrichment analysis with evoGO

Gene Ontology (GO) enrichment analysis is a powerful tool for elucidating underlying biological processes in high-throughput transcriptomic and proteomic studies. However, the redundancy of identified enriched GO terms, caused by the hierarchical nature of GO, significantly complicates the prioritization of relevant terms and makes drawing concise conclusions challenging. To address this problem, we developed evoGO, a novel method that aims to improve the specificity and relevance of enrichment results by considering the GO hierarchy during the analysis. Built on the foundation of conventional overrepresentation analysis (ORA), evoGO reduces the impact of differentially expressed genes on the significance of a given GO term if those genes already contribute to a higher significance of any descendant GO term. The effectiveness of the algorithm was evaluated against other advanced ORA-based GO enrichment analysis methods (topGO, clusterProfiler, and SetRank) using synthetic and real-life datasets. In the synthetic benchmarks, evoGO reduced the number of enriched terms identified by ORA on average by 30% while recovering the highest fraction (96%) of the true positives. When applied to real-life data, evoGO most efficiently prioritized tissue-specific GO terms in the analyses aimed to capture the biological processes inherent to various human tissues. Furthermore, evoGO eliminated the fewest enriched GO terms that had no significantly enriched relatives and thus should not be considered redundant. Finally, evoGO was one of the fastest of the tested methods and, unlike competitors, maintained consistent execution times across both benchmarks. In conclusion, our findings demonstrate that among tested ORA-based approaches, evoGO stands out as an effective and fast method for minimizing redundancy in GO enrichment analysis results while reliably preserving biologically relevant information. We believe that evoGO could be a valuable tool for conducting downstream analyses, particularly in the context of high-throughput transcriptomic and proteomic screening studies. The evoGO method has been implemented as an R package, which is available on GitHub.

bioinformatics↗

bioRxiv2022

Lung endothelium instructs dormancy of susceptible metastatic tumour cells

During metastasis, cancer cells hijack blood vessels and travel via the circulation to colonize distant sites1,2. Due to the rarity of these events, the immediate cell fate decisions of arrested circulating tumour cells (aCTC) are poorly understood and the role of the endothelium, as the interface of dissemination, remains elusive3,4. Here, we developed a novel strategy to specifically enrich for aCTC subpopulations capturing all cell states of the extravasation process and, in combination with single cell RNA-sequencing, provide a first blueprint of the transcriptional basis of early aCTC decisions. Upon their arrest at the metastatic site, tumour cells either started proliferating intravascularly or extravasated and preferably reached a state of quiescence. Endothelial-derived angiocrine Wnt factors were found to drive this bifurcation by inducing a mesenchymal-like phenotype in aCTCs instructing them to follow the extravasation-dormancy branch. Surprisingly, homogenous tumour cell pools showed an unexpected baseline heterogeneity in Wnt signalling activity and epithelial-to-mesenchyme-transition (EMT) states. This heterogeneity was established at the epigenetic level and served as the driving force of aCTC behaviour. Hypomethylation enabled high baseline Wnt and EMT activity in tumour cells leading them to preferably follow the extravasation-dormancy route, whereas methylated tumour cells had low activity and proliferated intravascularly. The data identify the pre-determined methylation status of disseminated tumour cells as a key regulator of aCTC behaviour in the metastatic niche. While metastatic niche-derived factors per default instruct the acquisition of quiescence, aCTCs unwind a default proliferation program and only deviate from it if hypomethylation in key gene families renders them responsive towards the microenvironment.

cancer biology↗

bioRxiv2019

Dissecting intratumor heterogeneity of nodal B cell lymphomas on the transcriptional, genetic, and drug response level

Tumor heterogeneity encompasses both the malignant cells and their microenvironment. While heterogeneity between individual patients is well-known to affect the efficacy of anti-cancer drugs, most personalized treatment approaches do not account for intratumor heterogeneity. We addressed this issue by studying the heterogeneity of lymph node-derived B cell non-Hodgkin lymphoma (B-NHL) by single cell RNA-sequencing (scRNA-seq) and transcriptome-informed flow cytometry. We identified transcriptionally distinct malignant subclones and compared their drug response and genomic profiles. Malignant subclones of the same patient responded strikingly different to anti-cancer drugs ex vivo, which recapitulated subclone-specific drug sensitivity during in vivo treatment. Tumor infiltrating T cells represented the majority of non-malignant cells, whose gene expression signatures were similar across all donors, whereas the frequencies of T cell subsets varied significantly between the donors. Our data provide new insights into the heterogeneity of B-NHL and highlight the relevance of intratumor heterogeneity for personalized cancer therapies.

cancer biology↗

bioRxiv2018

On the optimal design of metabolic RNA labeling experiments.

Massively parallel RNA sequencing (RNA-seq) in combination with metabolic labeling has become the de facto standard approach to study alterations in RNA transcription, processing or decay. Regardless of advances in the experimental protocols and techniques, every experimentalist needs to specify the key aspects of experimental design: For example, which protocol should be used (biochemical separation vs. nucleotide conversion) and what is the optimal labeling time? In this work, we provide approximate answers to these questions using asymptotic theory of optimal design. Specifically, we derive the optimal labeling time for any given degradation rate and show that sub-optimal time points yield better rate estimates if they precede the optimal time point. Subsequently, we show that an increase in sample numbers should be preferred over an increase in sequencing depth. Lastly, we provide some guidance on use cases when laborious biochemical separation outcompetes recent nucleotide conversion based methods (such as SLAMseq).

bioinformatics↗