MACHINE LEARNING IN GENOMIC DATA ANALYSIS

ML is a branch of artificial intelligence (AI) that involves enabling computers to make decisions from data without being directly programmed. It helps to process large amounts of complex data, identify intricate patterns, and develop predictive models that can guide personalized treatment and public health approaches [98]. It also plays an increasingly pivotal role in the analysis of extensive genomic datasets, offering the capability to discern patterns, predict treatment responses, and uncover novel therapeutic targets [99]. The goal of the ML approach is to acquire knowledge from historical or current data and use that understanding to make forecasts or decisions for unspecified future data metrics [100]. In genomic research, ML algorithms are widely applied across various domains. A well-established use case is genome annotation and predicting the effects of genetic variations, where ML techniques have become essential in bioinformatics processes [101]. These algorithms contribute significantly to enhancing our understanding of genomic variations and their potential functional consequences. The impact of ML extends into disease diagnosis and prognosis, where these algorithms adeptly analyze genomic data to identify patterns, enabling accurate disease diagnosis, prediction of disease progression, and informed treatment decisions [102].

Moreover, in liquid biopsy analysis, ML techniques prove valuable in scrutinizing data from various sources like CTCs and DNA, providing insights into cancer progression and treatment response [103]. ML techniques effectively identify disease-causing genomic variants by differentiating between pathogenic and benign variants. This significantly improves the accuracy of genetic diagnoses and enables more targeted treatment strategies for genetic disorders [104]. ML, particularly deep learning, contributes to improving gene editing tools like clustered regularly interspaced short palindromic repeats (CRISPR), enhancing precision in gene therapy strategies [105, 106]. Anticipating virus evolution, such as influenza and Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), is made more effective with ML, aiding public health efforts to stay ahead of outbreaks and formulate timely interventions [107]. Furthermore, ML algorithms play a crucial role in variant calling, i.e., the computational process of detection and annotation, enhancing the characterization of genetic variants in genomic data. These techniques reveal intricate relationships between gene expression levels and disease outcomes, furnishing researchers with invaluable insights that may hold the keys to unlocking novel therapeutic targets [108, 109]. In the vast field of drug discovery and development, the synergy between machine learning and large datasets is transforming the landscape [110]. ML algorithms facilitate the identification of potential drug targets, streamline the optimization of drug candidates, and even predict clinical trial outcomes, ushering in a new realm of efficiency and efficacy in pharmaceutical research [111]. Due to this, the field of personalized medicine has been greatly impacted by the excellence of ML in analyzing genomic data. This analytical capability enables the development of customized treatment plans for cancer patients, harnessing the power of individual genetic profiles to predict treatment responses with high precision and efficacy [112]. This highlights the significant and growing impact of ML on genomic data analysis, as researchers continually refine algorithms to extract valuable insights from the complexity of genomic datasets [113].

Multi-Omics Integration in Cancer Research

The word “omics” refers to the study of molecules in a biological sample. The word “omics” encompasses a wide range of studies, including genomics, proteomics, metabolomics, transcriptomics, and others [114-116]. Nowadays, omics approaches have gained various novel advances in cancer research. In oncology studies, it is widely recognized that cancer involves unavoidable molecular changes at different levels of biological organization. Omics technologies have thus facilitated a deeper exploration of cancer-related processes, allowing scientists to unravel its unique molecular characteristics [117].

Genomics is one of the most basic levels among the omics branches. In the field of oncology, the approach is known as cancer genomics [118]. Cancer genomics is the study of genetic changes associated with cancer development and progression, which helps to promote personalized cancer therapy [119]. Since cancer is largely caused by changes and mutations at the genome level, which help identify cancer-specific molecular signatures and the underlying mechanisms, advances in genomic-level research and related technologies aid in cancer management, from detection to treatment [120]. In this regard, DNA sequencing methods, from the first generation to next-generation sequencing (NGS), have unlocked new avenues for understanding the genetic codes of organisms, advancing research in the field [121]. Large-scale initiatives, such as the Human Genome Project, HapMap, and genome-wide association analysis [122], can be seen as advances in human genome research. Additionally, in cancer research, worthwhile projects such as TCGA, ICGC, and the Cancer Genome Project have been developed by analyzing DNA sequences to focus on cancer assessment in individuals [118].

Significant advancements have been made since the discovery of genomics, particularly in the field of cancer research. One of the major advantages in this field has been the identification of oncogenes (such as the KRAS family proto-oncogenes [123], EGFR [124], and the phosphatidylinositol-3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) signaling pathway [125]). The precise identification of the role of these oncogenes in the carcinogenic process has led to their recognition as suitable drug targets for cancer treatment [125].