• Dana Pe'er, Sloan Kettering Institute, United States

• Sofia Rodriguez, P.hD Student, Chile
• Camilo Villaman, Programa de Doctorado en Genómica Integrativa, Universidad Mayor, Chile
• Mauricio Saez, Facultad de ciencias de la salud. Departamento de procesos diagnósticos y evaluación, Universidad de Talca, Chile
• Alberto J. Martin, Facultad de Ingeniería, Arquitectura y Diseño, Universidad San Sebastián, Chile

Rett syndrome is a mono-chromosomal disorder with a prevalence much higher in women (95% of all cases). This disease is characterized by difficulties to study its phenotype caused by an intrinsic heterogeneity of the brain tissues affected due to the stochastic silencing of the X chromosome. In addition, we are yet largely unaware of the cellular-autonomous and non-cellular autonomous alterations that occur in neuronal populations because of this pathology. To try to overcome these issues, we performed a gene regulatory analysis, from human organoid single cell transcriptomic data. We first performed a trajectory analysis to understand the data characteristics, followed by generation of pseudo-time-based gene regulatory networks to assess non-cellular autonomous processes. These approaches made possible to show differences in the distribution of cells and their developmental differences. From the regulatory networks, it was possible to identify the absence of a regulatory mechanism associated with non-coding transcription factors not previously described for Dopaminergic neurons and a System of Master Regulators associated to MECP2 in Gabaergic neurons. These results served to postulate the existence of systemic master regulators, which we further evaluated with dynamic Boolean models per cell type based on the proposed regulatory network.

• Wisdom A Akurugu, University of Cape Town, South Africa
• Ekkehard Zöllner, Paediatric Endocrine Unit, Department of Pediatrics and Child Health, Stellenbosch University, South Africa
• Carel van Heerden, Stellenbosch University, South Africa
• Nicola Mulder, University of Cape Town, South Africa

Inhaled corticosteroids are crucial for managing asthma, but these may cause hypothalamic-pituitary-adrenal suppression (HPAS). Cortisol production is suppressed as it is in Addison’s Disease (AD), Clinical symptoms may be similar. Identifying genetic markers associated with ACTH resistance versus HPAS is crucial for precise patient stratification and personalized treatment. SNP data of ninety-six asthmatic children on inhaled corticosteroids and nasal steroids were studied. The participants underwent an overnight metyrapone test. Baseline adrenocorticotropic hormone and cortisol were measured as well as post-metyrapone adrenocorticotropic hormone (PMACTH). ACTH resistance was diagnosed if the PMACTH/C ratio is greater than 0.35. Eight-two and seventy-six samples out of the 96 participants had data for Basal and PM ACTH/C ratios respectively. SNP association was done using the PLINK analysis toolkit and statistical regression. Genetic models were assessed, and SNP functional annotation & prioritization were performed. Two significant SNPs emerged among others: rs6962 (G>A) on the SDHA gene and rs2303223 (G>A) on ZNF668 whereby rs6962 is likely resistant while rs2303223 is likely responsive to ACTH. Genotypic comparisons of rs6962 (AA vs GA & GG) and rs2303223 (GG vs GA & AA) showed statistical significance for √BasalACTH/C and √PMACTH/C ratios respectively. SDHA, integral to mitochondrial respiration, may affect metabolic pathways governing ACTH resistance. ZNF668, encodes a zinc finger protein and akin to other zinc proteins which are known to regular the glucocorticoid receptor (GR), may as well regulate the GR or other steroid receptors. Dysregulation of these genes could impact ACTH sensitivity. Further validatory and confirmatory investigations are recommended.

• Ornit Nahman, Technion – Israel Institute of Technology, Haifa, Israel, Israel
• Tim J. Cooper, Technion – Israel Institute of Technology, Haifa, Israel, Israel
• Shai S. Shen-Orr, Technion – Israel Institute of Technology, Haifa, Israel, Israel

Spatial transcriptomics (ST), a breakthrough technology, captures the complex structure and state of tissues. Several ST technologies now exist, most prominently spot-based platforms such as Visium. Despite ST's widespread usage and distinct data characteristics, the vast majority of studies continue to analyze ST data using algorithms built for older technologies, such as single cell (SC) and bulk RNA-seq. This is particularly the case when identifying differentially expressed genes (DEGs), however, it remains unclear if the approaches used are still valid for ST data. Here, we sought to characterize the performance of these methods by constructing an in-silico simulator of ST data with a controllable and known DEG ground truth. Surprisingly, our findings reveal little variation in the performance of classic DEG algorithms - all of which fail to accurately recapture known DEGs to significant levels. Importantly, we further demonstrate that cellular heterogeneity within spots is a primary cause of this poor performance and propose a simple gene-selection scheme, based on prior knowledge of cell-type specificity, to overcome this. Importantly, our approach outperforms existing data-driven methods and enhances DEG recovery and trustworthiness in ST data. Overall, our work details a conceptual framework that can be used upstream of any DEG algorithm to improve the accuracy of the results and of any subsequent downstream analysis.

• Javier Correa Alvarez, EAFIT, Colombia
• Juan Picon Cossio, EAFIT, Colombia
• Andres Florian Cruz, EAFIT, Colombia
• Fabian Uhrlaub, Hochschule Bremerhaven, Germany
• Susana Sierra Pelaez, Heinrich Heine Universität, Germany
• Carsten Harms, Hochschule Bremerhaven, Germany

Contamination of crops soils by cadmium (Cd) is a worldwide threat to ecosystems and human health. High concentrations of Cd damage the cell membrane, organelles, and generate overproduction of reactive oxygen species. Talaromyces santanderensis, a recently fungal strain isolated from cocoa soil, is a new species of Cd resistant fungus with a high tolerance rating between 100–400 mg/kg. However, there is no molecular or genetic information about the mechanism used for tolerance. Thus, this study presents the first draft genome assembly and annotation of T. santanderensis. The genome of this strain was sequenced using Oxford Nanopore Technology with fungal DNA under stress conditions of Cd at 50 ppm, and under normal conditions of growth. The genome was assembled using Flye, Miniasm, NECAT, Wtdbg2 and Canu and produced assemblies exhibiting similarly high levels of BUSCO completeness (~96.5%) with a coverage of 20x on average. Flye presented the assembly with the highest contiguity featuring a N50 length of 8,443,216 bp and a maximum contig of 12,809,360 bp. The genome we present has a GC content of 45.16% and a size of 38,889,036 bp. Gene prediction yielded 13,791 genes, and through functional annotation we identified important homologous protein-coding genes reported to be associated with Cd resistance such as copper chaperone, arsenite and zinc transporter, arsenical resistance protein, and Cd2+-exporting ATPase. Additionally, we present T. santanderensis methylome for the modified bases 5mC and 5mhC during both conditions, which will give us information about the role of the epigenomic in adaptation to heavy metal environment.

• Samuel Davis, School of Chemistry and Molecular Biosciences, The University of Queensland, Australia, Australia
• Gabriel Foley, School of Chemistry and Molecular Biosciences, The University of Queensland, Australia, Australia
• Marc Morris, School of Chemistry and Molecular Biosciences, The University of Queensland, Australia, Australia
• Mikael Bodén, School of Chemistry and Molecular Biosciences, The University of Queensland, Australia, Australia

The ability to confidently infer evolutionary relationships at the scale of protein superfamilies would profoundly transform biology. While the advent of various machine learning-based models for sequence analysis have provided significant improvements in remote homology detection, it is unclear whether such methods can effectively reconstruct phylogenies at this scale.

When analysing protein superfamilies, poor alignment accuracy presents a major hurdle to traditional phylogenetics. Additionally, the computational burden of performing these analyses on sufficiently large datasets may be prohibitive. These challenges are commonly exacerbated by input sequences which are poorly representative of the evolutionary space of interest. Curation approaches which rely on sequence similarity often produce unrepresentative sequences for alignment due to the diminishing correlation of sequence identity with evolutionary relatedness at high degrees of divergence.

We developed an approach to minimise these issues which constructs representative profile Hidden Markov Models (pHMMs) for sequence curation from large evolutionary spaces. From pHMMs built with robust alignments of highly similar sequences, the tool iteratively expands the profiles’ scopes. Their representativeness for the given space is optimised by systematic exclusion of sequence subsets and cross-validation over several iterations. Alignments and phylogenies constructed downstream of curation by this method demonstrate improvement across various metrics.

We applied this approach to investigate the evolutionary origins of B3 metallo-beta-lactamases, gaining novel insights into their emergence from a diverse superfamily. Enhancements in the alignment and phylogeny conferred by this tool allowed for characterisation of deep ancestral variants and new hypotheses regarding molecular determinants of this family’s function.

• Matthew T. Unger, UC Santa Barbara, United States
• Andrew P. Longhini, UC Santa Barbara Neuroscience Research Institute, United States
• Kenneth S. Kosik, UC Santa Barbara Neuroscience Research Institute, United States

Biomolecular condensates, such as stress granules (SG), are understood to harbor protein aggregates implicated in neurodegenerative pathologies like Alzheimer’s Disease (AD). Understanding biomolecular condensate dynamics may therefore provide insight into potential therapeutic interventions for these diseases. Biomolecular condensates are formed by liquid-liquid phase separation (LLPS). Under equilibrium conditions, multiple small droplets resulting from LLPS merge into a single large droplet. However, in living systems, phase-separated biomolecular condensates do not form large droplets and are instead maintained out-of-equilibrium as many small droplets. Molecular dynamics simulations run by our team suggested that this out-of-equilibrium behavior occurs due to an altered rate of switching between conformations of a biomolecule, with one state favoring phase separation and the other not. Based on this, we hypothesized that the faster a molecule switches between these two states, the further out-of-equilibrium and smaller the resulting biomolecular condensates will be. To investigate this, we used increasing cellular stress as a proxy for increasing swapping rate and explored the relationships between various stresses, ATP depletion, and SG sizes in a cell culture system. Additionally, we demonstrate that in vitro, phase-separated tau droplets remain smaller and more numerous when in the presence of the proline isomerase PPIA, compared to when PPIA is absent. By linking this PPIA to G3BP1, a core component of SGs, we showed that PPIA could control stress granule size. These findings together suggest that an out-of-equilibrium state can be maintained by an altered swapping rate of a biomolecule between two conformations.

• Jianlin Li, The University of Hong Kong, Hong Kong
• Qingqing Zhang, The University of Hong Kong, Hong Kong
• Cheuk-Lun Lee, The Hong Kong Polytechnic University, Hong Kong
• Philip C.N. Chiu, The University of Hong Kong, Hong Kong

Background: Preeclampsia (PE) is a common gestational disease affecting 2-5% of all pregnancies, with its aetiology associated with defective trophoblasts differentiation. Currently, there is no reliable approach for either prediction or treatment of PE. Recent research highlights the significance of premature trophoblast senescence as a new characteristic of PE. Nonetheless, the molecular interactions between trophoblast senescence and differentiation are poorly understood.

Methods: We analyzed aging marker expression in trophoblast stem cells differentiating into extravillous trophoblasts (EVTs), identifying an aging-related gene set (ARGS). Using CRISPR knockout, we verified ARGS roles in trophoblast differentiation. Since miRNA is stable and allows easy PE diagnosis, we examined interactions between ARGS and miRNA using covalent ligation of endogenous argonaute-bound RNAs-crosslinking and immunoprecipitation (CLEAR-CLIP).

Results: Our findings demonstrated that the EVT differentiation process encompasses a senescence process, characterized by elevated levels of ARGS. By employing CRISPR-screening, we demonstrated that 41 genes (TD-ARGs) in ARGS are involved in regulating trophoblast differentiation. 91 miRNAs (TD-ARG miRNA) that interact with TD-ARGs in trophoblast were then identified using our CLEAR-CLIP data. To test the relevance of these TD-ARG miRNA to PE, we examined their expressions in the serum of first-trimester pregnant women. The levels of 4 TD-ARG miRNA was significantly different in women who later developed PE when compared to those with normotensive pregnancy.

Conclusion: The results of this study provide novel evidence that senescence process is associated with trophoblast differentiation. Clinically, they also indicate the possible use of serum miRNA that targeting TD-ARG in the early prediction of PE.

• Iria Pose Lagoa, Barcelona Supercomputing Center - Universitat Politecnica de Catalunya, Spain
• José Carbonell-Caballero, BSC, Spain
• Beatriz Urda-García, Barcelona Supercomputing Center (BSC), Barcelona, Spain, Spain
• Jon Sánchez, Barcelona Supercomputing Center - Centro Nacional de Supercomputación, Spain
• Alfonso Valencia, Barcelona Supercomputing Centre BSC, Spain

Complex diseases often present a wide landscape of molecular profiles, posing challenges in identifying biomarkers associated with disease progression and, consequently, efficient personalized therapies. This is the case of Chronic Obstructive Pulmonary Disease (COPD), a heterogeneous condition characterized by the development of severe airflow obstruction profiles. Here, we propose a comprehensive framework comprising three key components: feature selection, prediction performance, and SHAP values analysis. To achieve this, we used gene expression data from the Lung Tissue Research Consortium and employed various feature selection criteria to identify the most relevant discriminant genes. These filtering approaches include knowledge extracted from intrinsic data characteristics (data-driven), external information from DisGeNET of genes associated with COPD (curated COPD-related genes), and their respective biological expansions based on physical interaction partners (OmniPath) and network-based prioritization algorithms (GUILDify). Subsequently, we exhaustively evaluate the performance of several state-of-the-art classifiers: Random Forest, Support Vector Machines - polynomial and radial kernel, k-Nearest Neighbors, Generalized Linear Models, and XGBoost. SHAP values obtained from the most accurate settings were combined with clustering algorithms to explain the model’s predictions and perform a cluster-based analysis for COPD profiling. Our integrative approach successfully distinguished COPD patients with data-driven selection strategy achieving the highest performance. These classifiers demonstrated accuracies of up to 84.8%, surpassing previous approaches. The genes identified in this study serve as promising biomarkers for COPD subtyping, offering avenues for more personalized treatment modalities. This study underscores the potential of integrated analytical approaches in advancing the diagnosis and treatment of complex diseases such as COPD.

• Manoj M Wagle, The University of Sydney, Children's Medical Research Institute, Australia
• Chunlei Liu, The University of Sydney, Children's Medical Research Institute, Australia
• Ellis Patrick, The University of Sydney, Australia
• Pengyi Yang, The University of Sydney, Children's Medical Research Institute, Australia

Single-cell omics technologies, with their recent advancement towards multi-modality, have achieved remarkable success in uncovering cellular heterogeneity at an unparalleled resolution. However, the high dimensionality, inherent noise, and sparsity render feature selection a crucial step in analyzing such data. Currently, the popular approach has been identifying highly variable genes, but this might not capture the complete spectrum of molecular variability and can miss out on certain informative genes. Moreover, the number of tools adept at capturing valuable information embedded in other single-cell modalities, such as chromatin accessibility and surface proteins, is currently limited.

To bridge this gap, we have developed Hydra, an interpretable deep generative framework based on variational autoencoders and data augmentation. Unlike traditional methods, Hydra is capable of effectively utilizing diverse single-cell omics data to capture cell-type specific molecular signatures, enabling a holistic examination of cells within the dataset. Additionally, as an integral component of this framework, we developed an ensemble classification module for the automated annotation of single-cell datasets. We extensively benchmarked Hydra across 23 datasets, including unimodal and multimodal single-cell omics datasets. Our results demonstrate that cell-type specific features selected by Hydra provide comparable to superior performance against several state-of-the-art methods in terms of stability, marker identification and cell-type annotation.

• Miriam Poley-Gil, Computational Biology Group, Life Sciences Department, Barcelona Supercomputing Center (BSC-CNS), Spain
• Maria I. Freiberger, Department of Biological Chemistry, Universidad de Buenos Aires (UBA), Argentina
• Alin Banka, Department of Informatics, Bioinformatics & Computational Biology, Technical University of Munich (TUM), Germany
• Michael Heinzinger, Department of Informatics, Bioinformatics & Computational Biology, Technical University of Munich (TUM), Germany
• Noelia Ferruz, Artificial Intelligence for Protein Design Group, Institute of Molecular Biology of Barcelona (IBMB-CSIC), Spain
• Alfonso Valencia, Computational Biology Group, Life Sciences Department, Barcelona Supercomputing Center (BSC-CNS), Spain
• R. Gonzalo Parra, Computational Biology Group, Life Sciences Department, Barcelona Supercomputing Center (BSC-CNS), Spain

Recently, Deep Learning models have revolutionised the Molecular Biology field allowing us to explore the intricate interplay between protein sequence, structure and function faster. To understand what they are capturing and generating we have combined state-of-the-art protein models for inverse folding (such as ProstT5[1] and ProteinMPNN[2]) and for sequence generation (such as ProtGPT2[3] and ZymCTRL[4]) with biophysical analyses (Figure 1).
We have studied conservation patterns of local energetic frustration in artificial datasets to shed light on the evolutionary processes leading to the diversification of some protein families, under the assumption that proteins are optimised for folding and stability, but also evolutionarily selected to function. We have developed a tool called FrustraEvo[5] that measures such conservation within and between protein families (available in full on the server https://frustraevo.qb.fcen.uba.ar/).
We found that most of the highly frustrated native residues are related to functional aspects. These functional residues are mostly recovered by sequence generation models, suggesting that there are alternative ways to design proteins instead of the way explored by evolution. In the case of catalytic sites, they are also recovered by inverse folding models. We therefore point out a selective memory concerning functionality (primary level of memory (local)). However, ProteinMPNN, also recovers the main network of frustrated contacts of the functional domains even suggesting a tertiary level of memory (contacts). Thus, our approach promises to effectively unravel the intricacies of protein family boundaries and explore design options for understanding protein evolution.

• Prisha Rai, High Technology High School, United States

Alzheimer's disease (AD), affecting over 50 million worldwide, is a progressive disorder characterized by Tau protein aggregation, leading to significant impairments. Current FDA-approved AD drugs target amyloid-beta aggregation, highlighting the need for alternative pathway approaches. This study introduces a novel approach integrating computational predictions with validations to identify therapeutic molecules against Tau aggregation. A variational autoencoder (VAE), using a Keras architecture, included a recurrent neural network (RNN) encoder and decoder and a property prediction model, with a gated recurrent unit (GRU) layer added to the decoder. The model was trained using the ZINC 250k Molecules Database and tested on a unique database of 72 known Tau inhibitors. The outputs synthesized Methylene Blue (MB), with validation loss sub-0.2 post 20 epochs. MB has been shown to inhibit Tau aggregation by accelerating LLPS. Administered to Alzheimer’s mutant D. melanogaster, MB increased RNA concentration in Tau-mutated flies’ brains. RNA quantification, a measure of transcriptional activity, validated MB's potential as a Tau inhibitor. This integrative approach highlights the efficacy of combining computational predictions with empirical testing in drug discovery.

• Manuel Corpas

• Wisdom A Akurugu, University of Cape Town, South Africa
• Ekkehard Zöllner, Paediatric Endocrine Unit, Department of Pediatrics and Child Health, Stellenbosch University, South Africa
• Carel van Heerden, Stellenbosch University, South Africa
• Nicola Mulder, University of Cape Town, South Africa

Inhaled corticosteroids are crucial for managing asthma, but these may cause hypothalamic-pituitary-adrenal suppression (HPAS). Cortisol production is suppressed as it is in Addison’s Disease (AD), Clinical symptoms may be similar. Identifying genetic markers associated with ACTH resistance versus HPAS is crucial for precise patient stratification and personalized treatment. SNP data of ninety-six asthmatic children on inhaled corticosteroids and nasal steroids were studied. The participants underwent an overnight metyrapone test. Baseline adrenocorticotropic hormone and cortisol were measured as well as post-metyrapone adrenocorticotropic hormone (PMACTH). ACTH resistance was diagnosed if the PMACTH/C ratio is greater than 0.35. Eight-two and seventy-six samples out of the 96 participants had data for Basal and PM ACTH/C ratios respectively. SNP association was done using the PLINK analysis toolkit and statistical regression. Genetic models were assessed, and SNP functional annotation & prioritization were performed. Two significant SNPs emerged among others: rs6962 (G>A) on the SDHA gene and rs2303223 (G>A) on ZNF668 whereby rs6962 is likely resistant while rs2303223 is likely responsive to ACTH. Genotypic comparisons of rs6962 (AA vs GA & GG) and rs2303223 (GG vs GA & AA) showed statistical significance for √BasalACTH/C and √PMACTH/C ratios respectively. SDHA, integral to mitochondrial respiration, may affect metabolic pathways governing ACTH resistance. ZNF668, encodes a zinc finger protein and akin to other zinc proteins which are known to regular the glucocorticoid receptor (GR), may as well regulate the GR or other steroid receptors. Dysregulation of these genes could impact ACTH sensitivity. Further validatory and confirmatory investigations are recommended.

• James W. Li, Department of Computer Science, Wake Forest University, Winston-Salem, NC, USA, United States
• Ravi Bandaru, Dept. of Biochemistry and Molecular Genetics, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA, United States
• Yaping Liu, Dept. of Biochemistry and Molecular Genetics, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA, United States

The fragmentation pattern of cell-free DNA (cfDNA) represents a promising non-invasive biomarker for disease diagnosis and prognosis. Numerous fragmentation features, such as end motif and window protection score (WPS), have been characterized in cfDNA genomic sequencing. However, the analytical tools developed in these studies are often not released to the liquid biopsy community or are inefficient for processing large datasets. To address this gap, we have developed FinaleToolkit, a fast and memory-efficient Python package designed to generate comprehensive fragmentation features from large cfDNA genomic sequencing data. For instance, FinaleToolkit can generate genome-wide WPS features from a ~100X cfDNA whole-genome sequencing dataset in 74 minutes using 16 CPU cores and 49 GB of memory, offering up to a ~30-fold increase in processing speed compared to original implementations. We have benchmarked FinaleToolkit against existing studies or implementations where possible, confirming its efficacy. Furthermore, FinaleToolkit is open source and thoroughly documented with both command line interface and Python application programming interface (API) to facilitate its widespread adoption and use within the research community: https://github.com/epifluidlab/FinaleToolkit

• Adeline McKie, Patrick G Johnston Centre for Cancer Research, Queen’s University Belfast; Belfast, Northern Ireland, United Kingdom
• Mark Wappett, Almac Discovery; Belfast, Northern Ireland, United Kingdom
• Benayu Priyanto, Patrick G Johnston Centre for Cancer Research, Queen’s University Belfast; Belfast, Northern Ireland, United Kingdom
• Hans Vandierendonck, School of Electronics, Electrical Engineering and Computer Science, Queen’s University Belfast; Belfast, United Kingdom
• Ian Overton, Patrick G Johnston Centre for Cancer Research, Queen’s University Belfast; Belfast, Northern Ireland, United Kingdom

Almost all Multiple Myeloma (MM) patients relapse and ultimately succumb to therapy-resistant disease; there is an urgent need for more effective treatment. Achilles' heel relationships arise when the status of one gene exposes a cell's vulnerability to perturbation of a second gene, such as chemical inhibition, providing opportunities for precision oncology.

We developed MultiSEp for integrative discovery of candidate gene dependency relationships in multiomics data. We predicted MM GDRs at genome-scale (27,232 genes, 370,777,296 candidate interactions) using transcriptomic data from the Multiple Myeloma Research Foundation (MMRF) CoMMpass study (n=928 patients). Filtering steps to derive a high-confidence synthetic lethal network (SynLethNet) included predicting characteristic mutual exclusive loss patterns (q<0.05). We predicted the population coverage achieved by drugging SynLethNet genes, and the impact of deleterious mutations. Our analysis only utilised deleterious mutations predicted by the variant effect prediction tools, SNPeff and SNPsift (annotated ‘high-impact’ mutations).

We characterised GDRs in the CoMMpass cohort and derived a high confidence predicted synthetic lethal network (1,466 genes, 5,245 edges; SynLethNet). Functional annotation of SynLethNet revealed many genes involved in the ubiquitin-proteasome system, which is dysregulated in MM and a target of current front-line therapy. Predictions were validated with the Cancer Dependency Map and the Cancer Therapeutics Response database.

We present the MultiSEp R package, demonstrated with a case study in Multiple Myeloma where we predict candidate drug targets and provide mechanistic insights to advance precision oncology.

• Sara Fumagalli, University of Milano-Bicocca, Italy
• Giulia Ghisleni, University of Milano-Bicocca, Italy
• Alice Armanni, University of Milano-Bicocca, Italy
• Luca Corneo, University of Milano-Bicocca, Italy
• Maurizio Casiraghi, University of Milano-Bicocca, Italy
• Antonia Bruno, University of Milano-Bicocca, Italy

Ensuring access to the bioinformatics field shall be on the agenda of life sciences degrees. This especially applies to biology-related degrees, where future biologists often ignore its existence due to the scarcity of dedicated courses.

We present our contribution to bioinformatics education, using the Data Hunters Workshop as a case study. Kicked off on February 28th, Data Hunters constitutes an ongoing student-science activity for students of the University of Milano-Bicocca, particularly those in the Biotechnology and Biosciences Department. Combining educational engagement with scientific research, this initiative enables students to tackle the key issue of metadata standardization’s lack in metagenomics, while supporting their learning. We provided a 6-hour lecture and an autonomous hands-on phase, structured as a learn-and-play activity with in-house built online educational and command-line resources. Thus, 29 students gained the fundamentals of metagenomics and Python language. Thanks to these tools, they have now stepped into the role of bioinformaticians, actively curating metadata from 379 amplicon-based and shotgun sequencing projects of the human skin microbiome to collaboratively create a curated metadata collection. Upon the workshop’s conclusion, we will assess the efficacy of our activity via surveys and standardized evaluation scales.

Our workshop represents the effort to bridge educational and research aims, including students as bioinformaticians of the future. As a reflection of the synergy of our objectives, the workshop's outcomes will include significant educational impacts that will lead to the development of a collaborative curated collection of human skin microbiome metadata, alongside the advancement of bioinformatics dissemination.

• Nilson Da Rocha Coimbra, Laboratorio de Bacteriologia, Faculdade de Medicina, Universidade de São Paulo, São Paulo, Brazil, Brazil
• Bernardo Velozo, Programa de Pós Graduação em Bioquímica, UFRJ, Rio de Janeiro, Brazil, Brazil
• Clara Carvalho, Programa Interunidades de Pós-graduação em Bioinformática, Universidade de São Paulo, São Paulo, Brazil, Brazil
• Lucas Aleixo Leal Pedroza, Programa de Pós-graduação em Biologia Aplicada à Saúde, Instituto Keizo Asami, UFPE, Recife, Brazil, Brazil
• Emerson Danzer, Programa Interunidades de Pós-graduação em Bioinformática, UFMG, Belo Horizonte, Minas Gerais, Brazil, Brazil
• Sandy Ingrid Aguiar Alves, Programa de Pós-Graduação em Biologia de Agentes Infecciosos e Parasitários, UFPA, Belém, Pará, Brazil, Brazil
• Rayssa Feitosa, The Hospital for Sick Children, Genetics and Genome Biology Department, Toronto, Canada, Canada
• Maira Neves, Programa Interunidades de Pós-graduação em Bioinformática, Universidade de São Paulo, São Paulo, Brazil, Brazil
• Bibiana Fam, Hospital de Clínicas de Porto Alegre, Porto Alegre, Rio Grande do Sul, Brazil, Brazil

The ISCB Regional Student Group of Brazil (RSG-Brazil) is at the forefront of promoting bioinformatics and computational biology education in Brazil. In 2019, RSG-Brazil launched its Educational Committee (EduComm) with a dual mission: to develop educational materials in Portuguese and to assess the efficacy of bioinformatics training in the country. Leveraging the ISCB core competency framework 3.0, the EduComm devised a novel training model to evaluate confidence in various technical aspects of bioinformatics. This model encompasses seven domain topics crucial for bioinformatics education: Biology, Statistics, Computer Science, Ethics, Bioinformatics Applications, Communication, and Professional Development. To gauge the educational needs of the Brazilian bioinformatics community, a survey was conducted in November 2023, collected 375 responses from across 21 states, predominantly from academia. Notably, the majority of respondents identified themselves as Bioinformatics Users, with a significant representation from Undergraduate and Graduate students. The survey revealed regional and profile-specific interests in Bioinformatics Topics, providing valuable insights for curriculum development. Through the efforts of EduComm, RSG-Brazil aims to tailor educational courses to meet the diverse needs of Brazil's computational biology student community. This work presents the findings from the survey conducted by RSG-Brazil's EduComm and highlights the importance of localized educational initiatives in advancing bioinformatics education on a global scale.

• Nils Gehlenborg
• DanDeBlasio DanDeBlasio
• Camilla-Castillo Vilchuaman

• Martin Steinegger

In metagenomics, DNA is sequenced directly from the environment, allowing us to study the vast majority of microbes that cannot be cultivated in vitro. This approach enables the real-time capture of pathogens, environmental monitoring, and access to a treasure trove of protein sequence diversity. However, annotating these metagenomes is particularly challenging, with many open reading frames remaining unannotated.

Advancements in protein structure prediction through methods like AlphaFold2 and ESMFold, have resulted in the AlphaFold databases and ESMatlas predicting over 214 and 620 million structures, respectively. In this talk, I will discuss how this avalanche of structural data can be used to improve genomic and proteomic annotation through rapid searches and clustering, and explore its potential to transform our understanding of microbial diversity.