Computational Molecular Biology 2024, Vol.14, No.3, 106-114 http://bioscipublisher.com/index.php/cmb 111 privacy is a major issue, as unauthorized access to genomic data can lead to misuse and exploitation (Azencott et al., 2018). Techniques such as fully homomorphic encryption (FHE) have been developed to enable the secure analysis of encrypted data, allowing for privacy-preserving applications in genomics without compromising the confidentiality of the data (Wood et al., 2020). Privacy-preserving AI techniques, including federated learning, have been proposed to protect individual privacy while enabling collaborative research (Torkzadehmahani et al., 2020). These methods aim to balance the need for data sharing in scientific research with the imperative to protect participant privacy. 6.2 Bias in machine learning models Bias in AI models is a critical ethical concern, particularly in the context of genomic research where biased data can lead to prejudiced outcomes. AI systems trained on biased datasets may produce results that disproportionately affect certain demographic groups, leading to issues of fairness and discrimination (Ntoutsi et al., 2020). For instance, AI tools developed using data from a homogenous population may not perform well when applied to diverse populations, exacerbating health disparities (Char, 2022). Addressing bias requires embedding ethical and legal principles in the design, training, and deployment of AI systems to ensure equitable outcomes. This includes deliberate efforts to minimize bias through diverse and representative data collection and rigorous validation processes. 6.3 Regulatory and legal issues The combination of AI and genomics introduces complex regulatory and legal challenges. The rapid advancement of these technologies often outpaces the development of regulatory frameworks, leading to uncertainties in their governance (Botes, 2023). The precautionary principle, which aims to prevent irreversible harm, has been suggested as a regulatory approach to manage the uncertainties associated with AI and genomics. There is a need for clear regulations to address data privacy, security, and ethical concerns in clinical AI systems (Gedefaw et al., 2023). The development of comprehensive and adaptive regulatory frameworks is essential to ensure the safe and ethical use of AI in genomic research, balancing innovation with the protection of individual rights and societal values. 7 Advances in AI Tools and Platforms for Genomic Research 7.1 Development of AI-driven genomic tools The integration of artificial intelligence (AI) into genomic research has led to the development of sophisticated tools that enhance the analysis and interpretation of complex genomic data. Machine learning (ML) applications have been particularly transformative, enabling the annotation of sequence elements and the analysis of epigenetic, proteomic, and metabolomic data. Platforms like PrismML exemplify this advancement by allowing users to perform multivariate machine learning on large genomic datasets, facilitating the identification of genotype-phenotype patterns and the prediction of clinical outcomes (Reddy et al., 2020). These tools leverage the computational power of cloud computing to handle the intensive processing requirements, making analyses faster and more scalable. 7.2 Cloud computing and big data integration The advent of cloud computing has revolutionized the way genomic data is processed and analyzed. Platforms such as the Genomics Virtual Laboratory (GVL) and Sherlock provide scalable, flexible, and accessible computational resources that are essential for handling the vast amounts of data generated by next-generation sequencing technologies (Afgan et al., 2015). These platforms offer a range of analysis and visualization tools, workflow management systems, and the ability to add or remove compute nodes as needed, thereby meeting the diverse demands of genomic researchers. The elasticity, reproducibility, and privacy features of cloud computing make it ideally suited for large-scale reanalysis of publicly available archived data, including privacy-protected datasets (Langmead and Nellore, 2018)。 7.3 Open-source platforms and collaboration Open-source platforms play a crucial role in fostering collaboration and innovation in genomic research. Tools like Sherlock and Machado provide comprehensive frameworks for storing, querying, and analyzing large
RkJQdWJsaXNoZXIy MjQ4ODYzNA==