• Alan Christoffels

Public health departments throughout the world have significant experience in the development and standardization of new laboratory techniques and protocols. This observation is underscored by the improvement of HIV laboratories between 2008-2018 in Sub-Saharan Africa that included at least 1100 laboratories of which 44 attained international accreditation. The COVID-19 pandemic has resulted in further strengthening of resources in public health institutes. But there has been and continue to be a need to strengthen computational skills and provision of bioinformatics tools to support public health decision making.

Over a period 5 years with strategic partners such as the Africa Center for Disease Control and Prevention we have supported the use of NGS technology in disease surveillance. A number of gaps were identified and resulted in projects to strengthen data governance and data management of priority pathogen data in Africa. As of 2019, we lead a global public health alliance for genomic epidemiology consortium to provide an opensource ecosystem that supports open bioinformatics software and data standards for public health utility. I will demonstrate some of these projects as part of our efforts to strengthen real-time decision making in public health disease outbreak response.

• Christian van Blerk, University of Pretoria, South Africa
• Teresa Coutinho, University of Pretoria, South Africa
• Pedro Lebre, University of Pretoria, South Africa

In this study the effects of Enterobacter ludwigii inoculation on onion bulbs were investigated by a disease progression analysis. Bulb rot in onions is a common occurrence in the storage of onions and leads to major losses in yield and money. Twenty four onions were inoculated with E. ludwigii and DNA extracted after 0 days (T0), 7 days (T1), 14 days (T2), 21 days (T3). The DNA was then sequenced using shotgun NGS sequencing. 23 samples were successfully sequenced. DNA filtering followed, with the onion host DNA removed so that only microbiome DNA remained. After this several Bioinformatic processes were performed with the goal of elucidating the bacterial component of the onion microbiome. This confirmed that we successfully established a new protocol for the extraction of microbial DNA from onion bulbs, and that we successfully inoculated the onions with E. ludwigii. Another goal was to establish whether predation occurred between the viral and bacterial components of our microbiome. How disease progresses in onions and plants is still not well understood, but through the use of bioinformatics and metagenomics we aim to expand our knowledge on this subject. This could lead to the establishment of new protocols to investigate these mechanisms of disease progression to aid in the development of potential biocontrol agents and the development of new curing and harvesting protocols.

• Iman Hamid, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Severine Nantenaina Stephie Raveloson, UNIVERSITY OF ANTANANARIVO, Madagascar
• Germain Jules Spiral, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Soanorolalao Ravelonjanahary, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Brigitte Marie Raharivololona, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• José Mahenina Randria, University of Antananarivo, Faculty of Medicine, Ministry of Public Health, Madagascar, Madagascar
• Mosa Zafimaro, University of Antananarivo, Faculty of Medicine, Ministry of Public Health,Madagascar, Madagascar
• Tsiorimanitra Aimée Randriambola, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Rota Mamimbahiny Andriantsoa, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Tojo Julio Andriamahefa, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Bodonomena Fitahiana Laza Rafidison, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Mehreen Mughal, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Anne-Katrin Emde, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Melissa Hendershott, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Sarah LeBaron von Baeyer, Division of Ethics & Engagement, Variant Bio Inc, United States
• Kaja A. Wasik, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Jean Freddy Ranaivoarisoa, University of Antananarivo, Mention Anthropobiologie et Développement Durable, Madagascar, Madagascar
• Laura Yerges-Armstrong, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Stephane E. Castel, Variant Bio, Inc., Seattle, WA 98109, USA, United States
• Rindra Rakotoarivony, Department of Anthropobiology and Sustainable Development, University of Antananarivo, Madagascar, Madagascar

The Malagasy population represents a unique case of human admixture, shaped by historical interactions between African- and Austronesian-ancestry groups. This complex demographic history has likely influenced patterns of genetic variation and trait evolution, yet the genetic basis of phenotypic diversity in this population remains understudied. Here, we investigate population structure and the genetic architecture of body composition using whole-genome sequencing(WGS).

We analyzed mid-pass WGS data from 214 Malagasy individuals across three localities across Madagascar, integrating anthropometric measurements to explore body composition variation. Population structure was inferred using principal component analysis (PCA) and ADMIXTURE to characterize African and Austronesian ancestry proportions. GWAS, conducted with Hail, identified genetic variants associated with body composition traits.

Our results reveal fine-scale variation in African and Austronesian ancestry across Madagascar, reflecting the population’s complex demographic history. GWAS identified novel variants influencing body composition, implicating genes involved in metabolic and skeletal pathways. These findings suggest that Malagasy populations harbor unique genetic variants, potentially shaped by natural selection, drift, and historical admixture events.

By combining WGS and GWAS, this study provides new insights into the evolutionary dynamics of an admixed population, shedding light on how genetic variation and demographic history contribute to phenotypic diversity. Our findings emphasize the importance of studying underrepresented populations to better understand human genetic diversity and evolution.

• Kimberly Coetzer, Department of Biomedical Sciences, Biomedical Research Institute, Faculty of Medicine and Health Sciences, South Africa
• A.S. Aidoo, Noguchi Memorial Institute for Medical Research, College of Health Sciences, University of Ghana, Legon, Accra, Ghana, Ghana
• N.A. Adomako, School of Medicine and Dentistry, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana, Ghana
• I.O. Ajiboye, CApIC-ACE, Covenant University, Ota, Nigeria, Nigeria
• H. Nortey, Noguchi Memorial Institute for Medical Research, College of Health Sciences, University of Ghana, Legon, Accra, Ghana, Nigeria
• O.I. Okello, Department of Immunology and Molecular Biology, School of Biomedical Sciences, Makerere University, Kampala, Uganda, Uganda
• O.I. Awe, African Society for Bioinformatics and Computational Biology, Cape Town, South Africa, South Africa

Introduction:
The reproducibility crisis in protein bioinformatics is a multifaceted issue stemming from several related challenges. A primary challenge is the inherent complexity and variability in proteomic datasets. Furthermore, the absence of standardized protocols and analysis pipelines across different laboratories and research groups compounds this problem, resulting in inconsistent practices and results. This review aimed to identify the primary obstacles and underlying causes contributing to the reproducibility crisis in proteomics analyses, while also exploring potential solutions.
Methodology:
The review introduced a novel approach to assess the reproducibility of proteomics tools. It evaluated factors like documentation, version control, maintenance and community engagement and rated them on a scale of one to three (greatest reproducibility). The Bio.tools database API was searched for keywords and tools were filtered in R using stringent criteria. Resulting tools were then scored based on the defined reproducibility criteria. Citation counts were analysed to assess the relationship between reproducibility metrics and daily application.
Result:
A total of 2379 tools were identified from the initial search. This was reduced to 107 tools following filtering. The results showed that 14 tools performed poorly, 42 tools had average reproducibility according to the criteria and 31 tools had good reproducibility according to the criteria (level 3). Most of the tools were for protein and peptide identification or data processing and analysis. Citation analysis showed that high reproducibility does not necessarily correlate with greater scientific value.
Conclusion:
By using this scoring system, developers can pinpoint areas requiring improvement and implement best practices to enhance reproducibility. Alternatively, users can select tools that are more likely to be reproducible. This will hopefully facilitate the development of community-driven standards and guidelines.
In conclusion, although the reproducibility crisis remains a challenge, continuous technological advancements will steadily mitigate its impact on research.

• Palesa Lesole, School of Molecular and Cell Biology, University of the Witwatersrand, South Africa
• Vanessa Meyer, School of Molecular and Cell Biology, University of the Witwatersrand, South Africa
• Nikki Gentle, School of Molecular and Cell Biology, University of the Witwatersrand, South Africa

Alternative splicing (AS) occurs in over 90% of multi-exonic genes, significantly increasing transcript and protein diversity. This tightly regulated process plays a crucial role in cell development and differentiation, and its dysregulation has been linked to various autoimmune diseases and cancers. Consequently, understanding AS is necessary to elucidate its role in normal cellular function and disease development. Advances in high throughput sequencing technologies, particularly RNA-seq, have led to the development of various tools and software packages for AS analysis.
AS analysis is broadly categorised into isoform-based and count-based methods. Isoform-based approaches rely on full transcript reconstruction and assess differential transcript expression between conditions. Count-based methods (further divided into exon- and event-level) provide a more straightforward approach to AS analysis and are better suited for short-read RNA-seq data. These approaches detect and quantify AS events and measure differential expression of transcript features (exons and junctions) between conditions by constructing genes as disjointed counting bins inferred from isoform-specific reads.
In this study, we employed three key count-based methods - differential exon usage (DEU), differential transcript usage (DTU) and differential AS events (DAS) - to investigate the global AS landscape during monocyte-to-macrophage differentiation. DEU was identified using DEXSeq (adjusted p-value < 0.05, |Log2FoldChange| > 1), DTU was identified using DRIMSeq (stage-wise analysis using stageR at overall FDR < 0.05), and DAS events were identified using rMATS-turbo (FDR < 0.05, |ΔPSI| > 0.1) in monocytic THP-1 cells and those treated with PMA for 96 hours to induce differentiation.
We identified 451 differentially used exons, 679 differentially used transcripts and 238 differential AS events affecting 160, 211 and 196 genes, respectively. Despite the relatively large number of AS events detected, there was limited overlap across methods, with only one gene detected by all three approaches. GO enrichment analysis using enrichGO revealed that DEU-associated genes were enriched in ribosomal development and ECM remodelling, whereas DTU-associated genes were enriched in GTPase regulator activity and negative regulation of cell killing. These findings suggest that each approach captures distinct aspects of AS regulation and contributes uniquely to understanding AS-driven changes in cellular differentiation.
The limited overlap between differentially spliced genes highlights the challenges of integrating count-based methods due to statistical and methodological differences among tools. However, these approaches provide unique and complementary insights at the exon, transcript and event levels, and uncover distinct regulatory mechanisms that shape cellular function during differentiation.

• Thandeka Malinga, Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand, South Africa
• Houcemeddine Othman, Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand, Tunisia
• Maria Paximadis, School of Molecular and Cell Biology, University of the Witwatersrand, South Africa
• Caroline Tiemessen, Centre for HIV and STIs, National Institute for Communicable Diseases, National Health Laboratory Services, South Africa
• Michèle Ramsay, Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand, South Africa
• Scott Hazelhurst, Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand, South Africa
• David Twesigomwe, Sydney Brenner Institute for Molecular Bioscience, University of the Witwatersrand, South Africa

Tuberculosis (TB) is a major health burden in Africa. Although TB is treatable, anti-TB drugs are associated with adverse drug reactions (ADRs), which are partly attributed to pharmacogenetic variation. The distribution of star alleles (haplotypes) influencing anti-TB drug metabolism is unknown in many African populations. This presents challenges in implementing genotype-guided therapy in Africa to decrease the occurrence of ADRs and enhance the efficacy of anti-TB drugs. In this study, we used StellarPGx to call variants and star alleles in NAT1, NAT2, GSTM1, GSTT1, GSTP1, and CYP2E1, from 1079 high-depth African whole genomes. We present the distribution of common, rare, and potential novel star alleles across various Sub-Saharan African (SSA) populations, in comparison with other global populations. NAT1*10 (53.6%), GSTT1*0 (65%), GSTM1*0 (48%), and NAT2*5 (17.5%) were among the predominant functionally relevant star alleles. Additionally, we predicted varying phenotype distributions for NAT1 and NAT2 (acetylation) and the glutathione-S-transferase (GST) enzymes (detoxification activity) between SSA and other global populations. Forty-seven potentially novel haplotypes were identified computationally across the genes. This study provides insight into the distribution of key variants and star alleles potentially relevant to anti-TB drug metabolism and other drugs prescribed across various African populations. The high number of potentially novel star alleles exemplifies the need for pharmacogenomics studies in the African context. Overall, our study provides a foundation for functional pharmacogenetic studies and potential implementation of pharmacogenetic testing in Africa to reduce the risk of ADRs related to treatment of TB and other diseases.

• Dougba Noel Dago and Jordan Aka

Cancers arise from genetic and epigenetic alterations, leading to uncontrolled cell growth. Gene fusion phenomena are recurrent in cancer cells. Breast cancer, a major public health concern, and several studies involved estrogens and estrogen nuclear receptors in its progression. Indeed, integration between genomic and bioinformatics tools strongly contribute in improving cancer molecular biology interpretation and comprehension. Herein, we performed a transcriptomic analysis aiming to assess gene fusion events in MCF7 breast cancer models that expressed estrogen nuclear receptors α/β under early (2h) estradiol (E2) stimulation. We aligned genomic reads sequences of that Breast Cancer model on the GRCh38 human genome, by using RNA-STAR. Next, we executed gene fusion calling via star-fusion package. Results showed a non-significant variability regarding gene fusion happening events between estradiol-stimulated and non-stimulated MCF7 Breast Cancer cells lines (p>0.05). Common detected genes fusions between analyzed Breast Cancer models result to be biomarkers of several cancers including breast cancer, and were characterized by the intra-chromosomic interactions. Findings revealed five (5) gene fusion events specific to breast cancer cells lines non-stimulated, and recognized as breast cancer biomarkers. Results exhibited estrogen nuclear receptor beta (Erβ) as inhibiting the expression of these Breast Cancer biomarkers in MCF7 breast cancer cells lines under estradiol stimulation. In conclusion, even if early estrogen hormone stimulation by inducing nuclear Erβ has non-significant impact on gene fusion variability between non-stimulated and stimulated breast cancer cell line models by contrast to the alternative splicing event, present study highlighted onco-suppressor activity of Erβ in breast cancer.

• Fatima Zahra Annassiri, ANISSE TEAM, Faculty of Sciences, Mohammed V University, Rabat, Morocco, Morocco
• Toheeb Jumah, School of Collective Intelligence, University Mohammed VI Polytechnic, Rabat, Morocco, Morocco
• Veronica Recheal Wokibula, South African National Bioinformatics Institute, University of the Western Cape, Cape Town, South Africa., South Africa
• Oussama Mohammed Reda, ANISSE TEAM, Faculty of Sciences, Mohammed V University, Rabat, Morocco, Morocco
• Olaitan I. Awe, African Society for Bioinformatics and Computational Biology, Cape Town, South Africa., South Africa

Introduction: Healthcare professionals face heavy documentation burdens, impacting patient care. This project addresses this challenge by developing a smart LLM, MediNote_Llama, for efficient clinical note generation.

Methods: We fine-tuned a Llama3 3B text-to-text model on the cleaned and anonymized PMC-Patients-V2 dataset. Our approach prioritizes efficient and accurate medical report generation with customizable output formats. The project aims to reduce documentation time, improve report accuracy and consistency, and demonstrate the potential of LLMs in healthcare automation.

Results: Performance was evaluated using BLEU, ROUGE, and medical domain accuracy metrics, achieving a comparable difference of 0.032 between the training loss and validation loss. We use a model to model summary comparison (GatorTron) to evaluate our model in the absence of human expert annotated.

Conclusion: The MediNote_Llama model is available on Hugging Face, enabling wider access and further development. This delivers a scalable system to free up valuable time for patient care.

• Blessing Sitabule, University of the Witwatersrand, South Africa
• Scott Hazelhurst, University of the Witwatersrand, South Africa
• Houcemeddine Othman, University of the Witwatersrand, Tunisia

Cytochrome P450 (CYP P450) enzymes are involved in over 90% of metabolic reactions that are known. To understand the interaction between CYP P450 enzymes and their ligands, a variety of experiments have been performed including in silico methods such as machine learning (ML) predictions. ML models can predict the potential interactions of enzymes with ligands of interest. In this study, we developed ML models based on several algorithms including the Support Vector Machine, Extreme Gradient Boosting (XGBoost), Random Forest, Neural-network and K-Nearest Neighbor using the Centre for High Performance Computing Lengau cluster. The models serve to predict the potency of ligands with respect to CYP P450 enzyme sequences. To develop the models, over 30 000 enzyme assay data was acquired from BindingDB. From the data, CYP P450 enzyme sequences were encoded using AAindex indices and the ligands were encoded using molecular descriptors, which were generated using RDkit and Mordred. In addition, the IC50 values of the enzyme assays were also encoded. Dimensionality reduction was performed on the enzyme and ligand features prior to training the models. The performance of the models was assessed using accuracy scores and the Area Under the Curve (AUC) of the Receiver of Operating Characteristic curves. The best performing algorithm was XGBoost with an accuracy of 0.8 and an AUC score of 0.9. This indicates that machine learning can be used to predict the potency of ligands with respect to enzymes of interest. Such models can potentially be employed for drug discovery, protein engineering and pharmacogenetics.

• Mohammed Khalifah, University of Hertfordshire Hosted by Global Academic Foundation, Egypt

Algae produce specialized metabolites (SMs) that aid adaptation to extreme environments and offer potential applications in pharmaceuticals, bioenergy, and biotechnology. However, their biosynthetic potential remains underexplored due to complex genomes and a lack of tailored bioinformatics tools. This study focuses on identifying biosynthetic gene clusters (BGCs) in 29 green algae species from the phylum Chlorophyta using genome mining techniques.

Publicly available genomic data from NCBI/JGI were analyzed using antiSMASH and other bioinformatics tools to predict and characterize BGCs. The results were filtered, validated, and visualized through clustering approaches to assess the distribution and diversity of biosynthetic pathways in different algae species.

The analysis revealed variations in the presence and abundance of secondary metabolite biosynthetic gene clusters among species. Notably, terpene biosynthetic gene clusters were prominent in Dunaliella salina and Haematococcus lacustris, likely associated with carotenoid production. NRPS and NRPS-like gene clusters suggested potential for bioactive peptide and polyketide biosynthesis. Some species, like Tetradesmus obliquus and Micromonas commoda, exhibited high BGC abundance, whereas others showed lower gene cluster expression.

This study highlights the untapped potential of algae as a source of valuable natural products. The findings emphasize the ecological significance and industrial potential of these metabolites while underlining the need for advanced genome mining tools for algal research. Future studies incorporating functional genomics approaches, such as gene knockout experiments, will help elucidate the roles of these biosynthetic pathways in algal adaptation and biotechnology applications.

• Heba Ali, Karolinska Institutet, Sweden
• Mukesh Varshney, Karolinska institutet, Sweden
• Per Nilsson, Karolinska Institutet, Sweden
• Ivan Nalvarte, Karolinska Institutet, Sweden

Objectives
Middle-aged women are 2-3 times more likely than men to develop Alzheimer’s disease (AD) later in life. While women tend to live longer, factors like sex hormones, genetics, and environment heighten their AD risk. Understanding sex differences in AD has been complex. This project aims to uncover the molecular underpinnings of the female sex hormone estrogen in AD models and relate these findings to human disease.
Methods
We are using APPNLGF mice, which exhibit clear AD pathology by 6 months of age, to study the effects of biological sex and sex hormones on memory and AD pathology at 6 and 12 months. Surgical menopause is induced in young adult female mice and castration in male mice. Preliminary data indicate neuroprotective effects of estrogen receptor beta (ERβ) in APPNLGF mice. Consequently, we crossed ERβ-/- mice with APP-KI mice to study the effects of ERβ loss on AD pathology, utilizing single-cell RNA sequencing of hippocampal brain cell populations.
Results
Our ongoing analysis provides insights into ERβ's role in brain cell functions and identifies sex-dependent alterations in gene activity. Preliminary data show that ERβ has neuroprotective effects in APPNLGF mice. We will anchor our findings to human disease by studying gene and pathway regulations in human AD brains and analyzing GWAS data from women with or without AD diagnosis and with or without different menopausal hormonal treatments.
Conclusions
Collectively, these findings offer insights into mechanisms underlying sex-specific susceptibility to AD and identify regulatory proteins for potential treatments targeting sex-dependent AD pathology. Understanding the role of estrogen and ERβ in AD could lead to targeted therapies addressing the heightened risk in women.

• Nomfundo Sikhakhane, Stellenbosch University, South Africa

Coronavirus Disease 2019 (COVID-19) is a respiratory illness caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). There are SARS-CoV-2 infected individuals with persisting severe disease that develop long-term complications, which is known as Long COVID. A possible outcome of a SARS-CoV-2 infection in younger individuals is called Multisystem Inflammatory Syndrome in Children (MIS-C), a rare hyperinflammatory condition. The inter-individual variability of clinical symptoms of COVID-19 is poorly understood. Host genetic factors have been proposed as a possible explanation with recent studies documenting complement dysregulation in SARS-CoV-2 infected individuals. Thus, the study aims to identify genetic variants within the complement system in patients with severe COVID-19, Long COVID and MIS-C using whole genome sequencing (WGS) and genotyping as well as assess the functional capability of the complement system in these individuals’ using immunoassays. Herein, we used WGS on 45 MIS-C, 20 severe COVID-19 and 20 Long COVID patients, and a customised SARS-CoV-2 genotyping array on 480 controls and COVID-19 patients. Following WGS and genotyping, variant calling files were generated then data quality control and principal component analyses were performed. Genome wide association studies are currently being performed to identify genetic variants within the complement system to establish the occurrence of specific variants associated with severe COVID-19, Long COVID, or MIS-C. Functional assessments will performed to measure the activity of the complement system in patients. The Complement CH50 Assay will be used to assess the total haemolytic activity of the complement system, and MILLIPLEX Complement Magnetic Bead Panel multiplex assay will quantify specific components of the complement system. The functional assessments will reveal the relative activity levels of these pathways during SARS-CoV-2 infection across severe COVID-19, MIS-C, and Long COVID patients.

• Peace Bassey Osim, University of Calabar Nigeria, Nigeria
• Blessing Bassey Ekpenyong, University of Calabar Nigeria, Nigeria
• Anita Yemi-Odae Nelson, University of Calabar Nigeria, Nigeria
• Bede Awan, University of Calabar, Nigeria, Nigeria
• Mary E. Kooffreh, University of Calabar Nigeria, Nigeria

Hypertension is a leading risk factor for cardiovascular diseases, with a disproportionately high prevalence in Africa. Angiotensin-converting enzyme (ACE) inhibitors are widely used to manage hypertension, but genetic variability, drug resistance, and accessibility issues necessitate the search for alternative treatments. Drug repurposing through in silico molecular docking offers a cost-effective approach to identifying new ACE inhibitors among FDA-approved drugs. In this study, 1,200 FDA-approved drugs were screened against ACE using molecular docking. The 3D structure of ACE was retrieved from the Protein Data Bank (PDB ID: 1O86), and docking simulations were performed using AutoDock Vina. Binding affinity scores were calculated to rank drug candidates, and the top 10 drugs with the highest binding affinities were selected for further analysis. The identified candidates included statins and antidiabetic agents, which demonstrated strong ACE inhibition potential. These findings highlight the potential of in silico drug repurposing in identifying novel antihypertensive candidates suitable for African populations, considering pharmacogenomic factors and healthcare accessibility challenges. Future experimental validation is necessary to confirm the therapeutic potential of these candidates in managing hypertension in African patients.

• Minenhle Mayisela, Department of Internal Medicine, The University of the Witwatersrand, South Africa
• Phelelani T. Mpangase, University of the Witwatersrand, South Africa
• Dineo Mpanya, The University of the Witwatersrand, South Africa
• Megan Shuey, Vanderbilt University Medical Centre, United States
• Quinn Wells, Vanderbilt University Medical centre, United States
• Roy Zent, Vanderbilt University Medical centre, United States
• Nqoba Tsabedze, The University of the Witwatersrand, South Africa

Accurate variant calling is essential for next-generation sequencing (NGS) data interpretation in diverse populations. While Illumina’s Emedgene platform provides automated variant interpretation, the Genome Analysis Toolkit (GATK) workflow offers a customizable, open-source alternative. This study compares variant calling outputs from Emedgene and GATK in a Southern African patient cohort to assess their concordance and implications for variant interpretation. Whole exome sequencing (WES) was performed on 101 patient samples from Southern Africa. Variants were identified using both Emedgene and a GATK-based pipeline. Filtering was applied based on quality metrics, allele frequency, and predicted pathogenicity. Concordance was evaluated by comparing the number, classification, and clinical relevance of detected variants. A total of seven variants were identified across the cohort. Six (85.7%) were concordant between the two workflows, while one variant was detected by GATK. The discordant variant may be attributed to differences in variant calling thresholds, filtering criteria, or annotation databases. Further investigation is needed to determine its clinical significance, particularly in the underrepresented Southern African genomic landscape. The high concordance between Emedgene and GATK suggests both pipelines are reliable for variant identification. However, discrepancies highlight the importance of cross-validation and careful interpretation when applying bioinformatics tools to genetically diverse populations. Understanding platform-specific methodologies and their impact on variant calling in underrepresented populations remains crucial for improving genomic analyses in Southern Africa.

• Alia Benkahla, Institut Pasteur de Tunis, Tunisia

I'll be presenting my career path over the last twenty years before I was able to start working in the field of systems biology, an area of research that is close to my heart. After a solid background in mathematics followed by a thesis in bioinformatics/genomics, I began my career with years of capacity building in bioinformatics in Africa and Tunisia. At the Institut Pasteur in Tunis, I set up a research group that became the Bioinformatics, Biomathematics and Biostatistics Laboratory (BIMS). It was a period of intense work, during which I trained students and led projects on crucial health issues for our region and the African continent as a whole. I was actively involved in initiatives such as the Tunisian node of H3ABioNet, and served for 12 years (from 2007 to 2019) on the Governing Council of the African Society for Bioinformatics and Computational Biology (ASBCB). The aim was to give African scientists the means to embrace computational biology projects. These experiences reinforced my belief in the power of quantitative approaches to solving biological problems and ultimately inspired my transition to the fascinating world of systems biology.