Biology subjects

Psomopoulos, F. E.

Publications and source records attributed to Psomopoulos, F. E..

5 recordsLinked to original sources

bioRxiv2025

lncAPNet enables the deciphering of lncRNA and mRNA connections in patient transcriptomic data

MotivationLong non-coding RNAs (lncRNAs) regulate gene expression through chromatin remodeling, transcriptional control, and post-transcriptional modulation, influencing physiological cell homeostasis but also disease onset. Yet most transcriptomic and network-based studies rely on descriptive linear co-expression analyses, missing nonlinear and mechanistic insights. Emerging ML/DL methods offer promise but remain limited by data sparsity, noise, insufficient biological priors, and poor interpretability, constraining systems-level lncRNA-mRNA motif discovery. ResultsIn this manuscript, we introduce lncAPNet, an extended version of APNet workflow, which integrates graph-based nonlinear inference of lncRNA-mRNA interactions using NetBID2s activity logic within an lncRNA-focused SJARACNe co-expression network, coupled with PASNet, a biologically informed sparse deep learning model. This framework enables explainable identification of lncRNA drivers in two different cancer type case studies, Chronic Lymphocytic Leukemia (CLL) and Prostate Adenocarcinoma (PRAD), uncovering lncRNA drivers that illuminate lncRNA-mediated programs in cancer progression. Availability and implementationlncAPNets R scripts, Python scripts, and methodologies are available at github repository: https://github.com/BiodataAnalysisGroup/lncAPNet

bioinformatics↗

bioRxiv2025

Mapping the AI Life Sciences Landscape in Greece: A National Survey and Bibliometric Comparison with Global Trends

Artificial intelligence is increasingly used in Life Sciences, though the pace and direction of adoption varies widely across countries. To map the Greek landscape, we combined two complementary approaches, a data-driven analysis of 916,824 AI-related life-science papers harvested from OpenAlex and PubMed and a targeted short survey at a national level aimed towards researchers, engineers, and support staff to be used as supporting material. We tagged each publication with Medical Subject Headings (MeSH) and compared topic frequencies between articles linked to at least one Greek institution and the rest of the world. Greek-affiliated outputs are disproportionately concentrated under the theme of methodology and algorithm-development, whereas the global corpus is dominated by disease-focused, organism-centered and clinical applications. Statistical contrasts across three MeSH hierarchy levels exposed clear national strengths in machine learning techniques and analytical tools, alongside under-representation in translational, patient-centred research. Survey responses reinforce these patterns: participants highlight limited access to well-curated biomedical data, constrained computational resources and difficulty recruiting cross-disciplinary talent as the chief barriers to progress. They advocate community-building actions (e.g., hackathons, postgraduate training, shared data infrastructures) that could realign national efforts with international practice while capitalizing on existing expertise in core-methodology development. Overall this study combines bibliometric evidence with community perspectives and provides a comprehensive overview of AI activity in Life Sciences in Greece, highlighting potential thematic strengths and gaps.

bioinformatics↗

bioRxiv2024

Synth4bench: a framework for generating synthetic genomics data for the evaluation of tumor-only somatic variant calling algorithms

MotivationSomatic variant calling is a key activity towards identifying genomic alterations; yet, the evaluation of the respective tools remains challenging due to the scarcity of high quality ground truth datasets. To overcome this limitation, we developed synth4bench, a synthetic data generation pipeline for robust benchmarking. Using a systematic process to create distinct synthetic datasets, we thoroughly evaluated five variant callers (Mutect2, FreeBayes, VarDict, VarScan2 and LoFreq). We compared tool outputs against our synthetic ground truth across key sequencing aspects (such as depth and read length) to assess their capacities and shed light on their underlying algorithmic principles. ResultsSynth4bench is an approach for evaluating tumor-only somatic variant callers that relies on a systematic definition of fully controlled ground-truth datasets. Our analysis revealed significant inconsistencies among the tool outputs and a strong dependence of caller performance on sequencing parameters. Indels remain the hardest-to-call variant type, driven by errors at low allele frequencies. Algorithmic choice is also critical; the most robust callers displayed the highest precision in allele frequency estimation, while the most sensitive caller was best for maximizing true positive recovery. Conversely, the least suitable caller exhibited systematic errors along with the poorest overall performance. These findings indicate that there isnt a one-solution-fit-all; sequencing optimization together with caller selection are necessary to maximize sensitivity and reliability. Furthermore, the pronounced inconsistencies suggest that current algorithms are not yet able to capture all mutational mechanisms adequately, with the modeling of the underlying processes remaining an open challenge. Availabilitycode: https://github.com/sfragkoul/synth4bench/ and data: https://zenodo.org/records/16524193 Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=69 SRC="FIGDIR/small/582313v2_ufig1.gif" ALT="Figure 1"> View larger version (16K): org.highwire.dtl.DTLVardef@1ac0fe5org.highwire.dtl.DTLVardef@1479cddorg.highwire.dtl.DTLVardef@8b7d1borg.highwire.dtl.DTLVardef@1c2a81b_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2022

Assessing SARS-CoV-2 evolution through the analysis of emerging mutations

IntroThe number of studies on SARS-CoV-2 published on a daily basis is constantly increasing, in an attempt to understand and address the challenges posed by the pandemic in a better way. Most of these studies also include a phylogeny of SARS-CoV-2 as background context, always taking into consideration the latest data in order to construct an updated tree. However, some of these studies have also revealed the difficulties of inferring a reliable phylogeny. [13] have shown that reliable phylogeny is an inherently complex task due to the large number of highly similar sequences, given the relatively low number of mutations evident in each sequence. MotivationFrom this viewpoint, there is indeed a challenge and an opportunity in identifying the evolutionary history of the SARS-CoV-2 virus, in order to assist the phylogenetic analysis process as well as support researchers in keeping track of the virus and the course of its characteristic mutations, and in finding patterns of the emerging mutations themselves and the interactions between them. The research question is formulated as follows: Detecting new patterns of co-occurring mutations beyond the strain-specific / strain-defining ones, in SARS-CoV-2 data, through the application of ML methods. AimGoing beyond the traditional phylogenetic approaches, we will be designing and implementing a clustering method that will effectively create a dendrogram of the involved sequences, based on a feature space defined on the present mutations, rather than the entire sequence. Ultimately, this ML method is tested out in sequences retrieved from public databases and validated using the available metadata as labels. The main goal of the project is to design, implement and evaluate a software that will automatically detect and cluster relevant mutations, that could potentially be used to identify trends in emerging variants. Contacttasos1109@gmail.com

bioinformatics↗

bioRxiv2022

Galaxy Training: A Powerful Framework for Teaching!

There is an ongoing explosion of scientific datasets being generated, brought on by recent technological advances in many areas of the natural sciences. As a result, the life sciences have become increasingly computational in nature, and bioinformatics has taken on a central role in research studies. However, basic computational skills, data analysis and stewardship are still rarely taught in life science educational programs [1], resulting in a skills gap in many of the researchers tasked with analysing these big datasets. In order to address this skills gap and empower researchers to perform their own data analyses, the Galaxy Training Network (GTN) has previously developed the Galaxy Training Platform (https://training.galaxyproject.org); an open access, community-driven framework for the collection of FAIR training materials for data analysis utilizing the user-friendly Galaxy framework as its primary data analysis platform [2]. Since its inception, this training platform has thrived, with the number of tutorials and contributors growing rapidly, and the range of topics extending beyond life sciences to include topics such as climatology, cheminformatics and machine learning. While initially aimed at supporting researchers directly, the GTN framework has proven to be an invaluable resource for educators as well. We have focused our efforts in recent years on adding increased support for this growing community of instructors. New features have been added to facilitate the use of the materials in a classroom setting, simplifying the contribution flow for new materials, and have added a set of train-the-trainer lessons. Here, we present the latest developments in the GTN project, aimed at facilitating the use of the Galaxy Training materials by educators, and its usage in different learning environments.

bioinformatics↗