Personalized therapy is a treatment strategy tailored to individual patients based on their genetic, phenotypic, and environmental characteristics. Compared to traditional disease treatment, personalized therapy is more precise and effective, significantly improving patient treatment response and survival rates. However, achieving personalized therapy requires an accurate understanding of the patient's disease characteristics and corresponding biomarkers.

In recent years, with the rapid development and substantial cost reduction of high-throughput sequencing technologies, we have entered an era of big data in bioinformatics. Technologies such as genomics, transcriptomics, proteomics, and metabolomics generate vast amounts of data, providing unprecedented opportunities for disease research and treatment (Kang et al., 2022). Against this transformative backdrop, personalized therapy has become a highly focused area.

The concept and methods of multi-omics data integration have become crucial tools for personalized therapy. Multi-omics data integration involves combining and analyzing multiple omics data sources from different levels to reveal the complex mechanisms of diseases and the specific responses of individuals. Comprehensive analysis not only uncovers deeper biological insights but also offers better personalized treatment plans for patients.

However, multi-omics data integration in personalized therapy still faces numerous challenges. Different types of multi-omics data have differences in data structure, scale, and accuracy, making effective integration and interpretation a key issue. Data quality and consistency are also challenges (Canzler et al., 2020), as data may be affected by sample preparation, measurement errors, and data missing. Additionally, data privacy and ethical issues must be addressed.

Therefore, this study introduces the applications and challenges of multi-omics data integration in personalized therapy. It discusses the sources and types of multi-omics data, introduces currently used data integration methods and tools, and explores the application fields of multi-omics data in personalized therapy. Furthermore, we will focus on the challenges and limitations of multi-omics data integration and propose future prospects and directions to promote the further development of personalized therapy, providing better medical services and efficacy for patients.

1 Sources and Types of Multi-omics Data

Multi-omics data encompass biological data from various levels, including genomics, transcriptomics, epigenomics, proteomics, and metabolomics. These data are typically obtained through high-throughput sequencing technologies and other high-throughput experimental methods.

1.1 Acquisition and application of genomics data

Genomics data are acquired through high-throughput sequencing technologies (Manni et al., 2021). After obtaining DNA samples, sequencing instruments are used to sequence them. Genome sequencing can produce a complete genome sequence of an individual, helping researchers understand the structure and function of the genome.

Genomics data have extensive applications in personalized therapy. Genomic data can help determine whether an individual carries genes associated with certain genetic disease risks, enabling early prevention and intervention. Genomic data can also be used for drug efficacy prediction. By analyzing an individual's genome sequence, it is possible to determine the efficacy and side effect risks of specific drugs for that individual, providing a basis for personalized drug therapy. Additionally, in cancer treatment, genomic data help identify driver gene mutations in tumors, providing guidance for selecting appropriate targeted therapy methods.

1.2 Acquisition and application of transcriptomics data

Transcriptomics data are a type of high-throughput experimental data that study gene expression levels within the genome. The acquisition and application of transcriptomics data provide researchers with important tools for understanding gene expression regulation and biological processes, and they are significant for biomedical research, disease diagnosis, and personalized therapy.

The acquisition and application of transcriptomics data generally involve the following steps. First, RNA is extracted from the tissues or cells to be studied (Liu et al., 2022). The extracted RNA can be total RNA, or specific types of RNA can be selected as needed, such as mRNA or non-coding RNA. Next, high-throughput sequencing technology (e.g., RNA-seq) is used to sequence the extracted RNA samples. RNA-seq can generate a large number of short-read sequences, providing information on gene expression levels and alternative splicing. For the sequenced data, data processing and analysis are required. This includes quality control, removing low-quality sequences, adapter trimming, and bioinformatics analyses such as aligning to the reference genome or transcriptome, quantifying gene expression, and differential expression analysis.

Transcriptomics data have a wide range of applications. By analyzing gene expression profiles, it is possible to understand the patterns and changes in gene expression in different tissues, developmental stages, and conditions, thereby revealing gene regulatory networks and potential biological functions. By comparing transcriptomics data from different samples, genes with significant expression differences in various biological processes, disease progression, and drug responses can be identified, leading to the discovery of potential biomarkers and drug targets. Furthermore, transcriptomics data can reveal alternative splicing events, and by integrating multiple transcriptome datasets, reconstruct the patterns of gene expression profiles and model gene regulatory networks. These analyses help in understanding gene regulatory mechanisms and identifying disease-associated splicing variants.

1.3 Acquisition and application of proteomics data

Proteomics is a discipline that studies the composition, structure, and function of proteins. The acquisition and application of proteomics data provide researchers with important tools to understand protein composition, structure, and function, as well as to reveal biological processes and disease mechanisms within organisms.

When conducting proteomics research, proteins must first be extracted from the biological samples under study. This typically involves steps such as cell lysis, protein extraction, purification, and concentration. Next, proteins can be separated using techniques such as two-dimensional gel electrophoresis or liquid chromatography. Two-dimensional gel electrophoresis separates proteins based on their molecular weight and isoelectric point, while liquid chromatography separates proteins based on their chemical properties. The separated proteins can then be identified and quantified using mass spectrometry techniques, such as tandem mass spectrometry (MS/MS) and time-of-flight mass spectrometry (TOF-MS). Mass spectrometry analysis measures the mass and amino acid sequence of proteins, thereby determining their identity. Finally, bioinformatics tools and databases are used to analyze and annotate the mass spectrometry data. This includes protein identification, modification site identification, protein interaction analysis, and functional annotation.

The applications of proteomics data are extensive. By analyzing proteomics data, researchers can understand the composition and abundance of proteins in cells or tissues, which in turn provides insights into the functions and physiological processes of biological systems. By comparing proteomics data from different samples, proteins with significant expression differences under different physiological or disease states can be identified, leading to the discovery of potential biomarkers or drug targets. Additionally, proteomics data can identify protein-protein interactions and functional associations, revealing protein regulatory networks, signaling pathways, and cellular functions. Proteomics data can also detect protein modifications such as phosphorylation, acetylation, and methylation, which are important for understanding protein regulation and function.

1.4 Acquisition and application of metabolomics data

Metabolomics is the study of the composition and changes of metabolites within organisms. The acquisition and application of metabolomics data provide researchers with important tools to understand the composition and changes of metabolites and their relationships with biological processes and diseases. Metabolomics has broad application prospects in fields such as medicine, biological sciences, agricultural sciences, and food sciences.

To acquire and apply metabolomics data, samples must be collected from biological specimens under study (e.g., blood, urine, tissues) and undergo preprocessing steps, such as protein removal and salt removal, to reduce the impact of interfering substances. Next, an appropriate analytical platform is selected for metabolite analysis. Common analytical techniques include gas chromatography-mass spectrometry (GC-MS) (Figure 1), liquid chromatography-mass spectrometry (LC-MS), nuclear magnetic resonance (NMR), and mass spectrometry imaging (MSI). Metabolite analysis is then performed on the samples to obtain high-throughput metabolomics data. These data provide information on the types, relative abundance, and changing trends of metabolites. Bioinformatics tools and statistical methods are used to process and analyze metabolomics data, including feature detection, mass spectrometry peak alignment, metabolite identification, quantification, and differential analysis.

Metabolomics data have widespread applications. By analyzing metabolomics data from different samples, metabolites associated with specific physiological states, diseases, or drug responses can be identified. These metabolites can serve as biomarkers (Wu et al., 2022), used for disease diagnosis, prognosis assessment, and monitoring treatment responses. By comparing metabolomics data from different samples, changes in metabolic pathways and key nodes in metabolic networks can be revealed. This helps in understanding the regulatory mechanisms of metabolism, the regulation of biological processes, and the mechanisms of disease occurrence. Metabolomics data can provide information about drug metabolism pathways in the body, drug metabolites, and pharmacokinetics. This is significant for drug development and personalized drug therapy. Metabolomics data can be used to analyze the effects of different foods or dietary patterns on metabolism, helping to understand the relationship between food and health and providing scientific evidence for personalized nutritional guidance.

2 Methods and Tools for Multi-omics Data Integration

2.1 Main methods of data integration

The main methods of data integration include clustering analysis and phenotypic clustering, association and network analysis, and machine learning and artificial intelligence methods. In practical applications, different methods and tools are often combined for multi-omics data integration and analysis to comprehensively understand the complexity and interrelationships of biological systems.

Clustering analysis is a method of grouping samples or features in a dataset based on their similarity. It can be used to cluster samples from different omics data to discover potential biological clustering patterns. Phenotypic clustering involves clustering phenotypic information (such as gene expression, protein expression, etc.) from multi-omics data and grouping samples with similar phenotypes into the same cluster. This helps in understanding the associations between different data types and their combined impact on biological characteristics.

Association analysis is used to find correlations or association patterns in multi-omics data. For example, correlation coefficients or mutual information metrics can be used to evaluate the correlation between different data and find related features. Network analysis methods consider biological molecules (such as genes, proteins, metabolites) in multi-omics data as nodes of a network, with edges representing their interactions. This helps to reveal regulatory mechanisms and biological associations in multi-omics data.

Machine learning and artificial intelligence methods can be used to integrate multi-omics data and discover patterns and rules in the data. For example, supervised learning algorithms (such as support vector machines, random forests) can be used for classification and prediction, while unsupervised learning algorithms (such as clustering, dimensionality reduction) can be used for data exploration and pattern discovery. Deep learning, a branch of machine learning, uses multi-layer neural network models to learn complex features and associations in multi-omics data. It has achieved great success in fields such as image recognition and natural language processing and is also applied to data integration and analysis in biology (Liu et al., 2022).

2.2 Common software tools and databases for data integration

There are several commonly used software tools and databases available for data integration. For databases and datasets, the NCBI database provides various biomedical data resources, the Ensembl database is a comprehensive genome annotation database, and the GEO database contains high-throughput transcriptomic data from around the world. Additionally, the TCGA database is a rich collection of cancer multi-omics data.

For data analysis and integration tools, R/Bioconductor is a popular language for statistical computing and data visualization, with packages such as limma, edgeR, DESeq2, and ConsensusClusterPlus available for multi-omics data analysis. Python is also widely used, with packages like pandas, numpy, scikit-learn, and TensorFlow suitable for data integration and analysis. Cytoscape is a powerful platform for network analysis and visualization, suitable for integrating multi-omics data and revealing relationships between biological molecules. The GSEA tool can be used to determine the enrichment of gene sets in different omics data. Additionally, Galaxy is a scientific workflow management system that can integrate and analyze multi-omics data (Afgan et al., 2018).

These tools and databases are just some of the common choices, and suitable software tools and databases can be selected based on specific research needs and data types in practical use.

3 Applications of Multi-omics Data Integration in Personalized Therapy

3.1 Disease prediction and diagnosis

One of the important applications of multi-omics data integration in personalized therapy is in disease prediction and diagnosis. By integrating data from various biomarkers, such as genomics, transcriptomics, proteomics, and metabolomics, a more comprehensive understanding of disease mechanisms can be achieved (Figure 2), providing more accurate information for personalized disease prediction and diagnosis.

By integrating multi-omics data, disease-related biomarkers can be identified, and predictive models can be developed to assess disease risk. For example, genomic data can be used to discover genetic variants associated with disease risk, transcriptomic data can reveal gene expression patterns, and proteomic data can identify protein biomarkers related to diseases. By combining these data, classifiers or risk assessment models can be constructed to help doctors determine a patient's likelihood of developing a disease and take preventive or early treatment measures.

Multi-omics data integration can provide more comprehensive and detailed disease diagnosis information. By measuring the genomics, transcriptomics, proteomics, etc., of patient samples and comparing them with known disease patterns in databases, the type and characteristics of the disease can be identified. For example, analyzing mutations, copy number variations, and chromosomal rearrangements in genomic data can classify and grade tumor types (Menyhárt and Győrffy, 2021). Additionally, combining transcriptomic and proteomic data can further define the molecular phenotype and drug sensitivity of the disease, providing a basis for precision treatment.

3.2 Drug development and screening

Another important application of multi-omics data integration in personalized therapy is in drug development and screening. By integrating biological data from different levels, the molecular mechanisms of diseases can be better understood, new drug targets can be discovered, and drug efficacy and side effects can be predicted, thereby accelerating the drug development and personalized drug screening process (Li et al., 2020).

Integrating data from genomics, transcriptomics, and proteomics can reveal key molecules and pathways involved in disease development. These data can be used to identify new drug targets, which are critical protein molecules involved in disease mechanisms. Further research on these potential targets can lead to the design of appropriate drug molecules to target diseases.

By integrating data from genomics, transcriptomics, and proteomics, models related to drug efficacy can be established. These models can predict individual responses to specific drugs, helping to determine the optimal treatment plan for patients. Analyzing an individual's genetic variations, gene expression patterns, and other biological characteristics can predict the patient's drug resistance, side effects, and treatment efficacy, providing guidance for personalized drug therapy.

Multi-omics data integration can also be used to screen and evaluate the efficacy and safety of potential drug molecules. By measuring genomics, transcriptomics, and proteomics data in disease model cell lines or animal models, the impact of drug molecules on disease characteristics and signaling pathways can be assessed. This helps to identify drug molecules with potential efficacy and lower side effects, improving drug development efficiency.

3.3 Treatment selection and response prediction

Another key application of multi-omics data integration in personalized therapy is in treatment selection and response prediction. By integrating biomarker data from patients, such as genomics, transcriptomics, proteomics, and metabolomics, doctors can determine the most suitable treatment plan for patients and predict their response to treatment (Subramanian et al., 2020).

Integrating biomarker data from patients can provide detailed information about disease molecular characteristics and signaling pathway activation states. This helps to understand individual differences in disease, including genetic variations, protein expression patterns, and metabolic characteristics. Based on this information, doctors can choose the best treatment plan tailored to the patient's specific disease molecular characteristics, improving treatment efficacy and survival rates.

By integrating biomarker data from patients, their response to different treatment plans can be predicted. For example, analyzing mutations and copy number variations in genomic data, gene expression patterns in transcriptomic data, and protein biomarkers in proteomic data can predict a patient's sensitivity, resistance, and side effects to specific drug treatments. This helps doctors choose the optimal treatment plan and provide personalized treatment recommendations, avoiding ineffective or toxic treatments.

Multi-omics data integration can also be used for dynamic monitoring and adjustment of treatments. By regularly measuring biomarker data from patients, the effectiveness of treatment can be promptly assessed, and treatment plans can be adjusted based on individualized data feedback. This ensures personalized and optimized treatment, improving treatment success rates and patient survival rates.

3.4 Personalized nutrition and lifestyle interventions

Another important application of multi-omics data integration in personalized therapy is in personalized nutrition and lifestyle interventions. An individual's genomic, transcriptomic, and metabolomic data can provide information about their metabolic state, nutritional needs, and responses. By integrating these data, tailored nutrition and lifestyle intervention plans can be designed to meet each individual's unique needs.

By analyzing an individual's genomic and metabolomic data, their ability to metabolize and absorb different nutrients can be understood. This helps identify which types of nutrients are most important for the individual's health and whether they have any metabolic deficiencies or intolerances to certain nutrients. Based on this information, personalized nutritional intervention plans can be designed to ensure adequate nutrient intake and prevent health issues caused by nutritional deficiencies.

Integrating an individual's genomic, transcriptomic, and environmental data can help understand their responses to different lifestyle factors. This includes their reactions to exercise, sleep, stress, and other lifestyle elements. Based on this information, personalized lifestyle intervention plans can be designed to meet their health needs. For example, increasing physical activity may be more effective for some individuals' health, while improving sleep quality may be more crucial for others. Personalized lifestyle interventions can improve an individual's health and prevent the onset of diseases.

Personalized nutrition and lifestyle interventions can also play a significant role in individuals who already have certain diseases. By integrating multi-omics data, vulnerabilities to specific nutrients and lifestyle factors, as well as potential health risks, can be identified. Based on this information, personalized health management plans can be developed, including dietary, exercise, and psychological interventions targeted at specific diseases to improve patient health.

4 Challenges and Limitations of Multi-omics Data Integration

4.1 Data quality and consistency issues

The quality of omics data is crucial for the accuracy and reliability of integration. The use of different laboratories, platforms, and technologies may lead to variations in data quality, such as measurement errors, systemic biases, and data loss. These issues can affect the reliability and comparability of the data, thereby impacting the accuracy of integration analysis and the interpretation of results (Menyhárt and Győrffy, 2021).

Multi-omics data often come from various sources, such as genomics, transcriptomics, proteomics, and metabolomics. These data may use different experimental designs, technical platforms, and analytical methods, leading to inconsistencies between datasets. These inconsistencies may include imbalances between different data levels, heterogeneity in data types, and inconsistencies in measurement units. These issues increase the complexity of data integration and affect the results of comprehensive analyses.

Ensuring the matching and consistency of samples in multi-omics data integration is crucial for interpreting results. For example, in the integration of genomic and transcriptomic data, it is important to ensure the consistency of genotype information with nucleic acid expression data. Additionally, sample selection and matching should consider biological variations, disease states, and treatment histories to avoid potential confounding effects and misleading results.

Multi-omics data typically exist in large-scale and high-dimensional forms, presenting challenges in data storage and management. Moreover, due to the diversity of data sources and the complexity of comprehensive analyses, effective data sharing and communication are also essential. Therefore, establishing appropriate data storage and sharing strategies, as well as promoting data exchange and collaboration, becomes a challenge.

4.2 Data missingness and incompleteness

Multi-omics data may contain missing data, where certain variables or observations are not recorded or collected. Missing data can result from limitations in experimental techniques, defects in experimental design, or issues in sample processing workflows. Missing data lead to information loss and reduced sample size, decreasing data reliability and interpretability.

Apart from missing data, multi-omics data may also have issues of data incompleteness, where information on certain variables or observations is incomplete or inaccurate. This can result from data recording errors, data processing errors, or other experimental or technical issues. Data incompleteness can cause inaccuracies in analysis and biases in results.

In the presence of missing and incomplete data, handling these issues becomes crucial. Common approaches include deleting missing data, imputing missing data, or using specialized algorithms for handling missing data. However, these methods come with their own limitations and assumptions and may impact analysis results. Therefore, choosing an appropriate method for handling missing data requires careful consideration and should align with the data characteristics and analysis objectives.

Missingness and incompleteness in multi-omics data can lead to dataset imbalance, where the quantity and quality of different data groups are inconsistent. This can affect the reliability of analysis and the interpretation of results, especially in cases involving classification or modeling. Thus, appropriate strategies are needed to address data imbalance, such as data resampling and adjustment.

4.3 Complexity and difficulty of data integration

Multi-omics data integration typically involves multiple data types, such as genomics, transcriptomics, proteomics, and metabolomics. These data types may have different characteristics and structures, requiring appropriate methods for integration. The complexity of data integration lies in how to merge information from different data types to obtain comprehensive and holistic insights.

Multi-omics data are usually high-dimensional and large-scale, involving numerous variables and samples. Handling and analyzing such large-scale data require efficient computational methods and sufficient computational resources. Additionally, high-dimensional data face the curse of dimensionality, where increased dimensions can lead to difficulties in modeling and interpretation.

Measurement methods and technical platforms from different data sources may lead to consistency issues between datasets. Moreover, data standardization is a significant challenge, as the same measurement indicators may have different units and scales across different data types and platforms. Ensuring consistency and standardization of data is a critical step in integrated analysis but requires appropriate data transformation and adjustment.

In data integration, selecting appropriate algorithms and models to handle and analyze complex multi-omics data is a key issue. Due to the diversity of data types and structures, new algorithms and models need to be developed to adapt to different data integration scenarios. Additionally, model selection must consider the data characteristics, the objectives of the problem, and the need for interpretability.

4.4 Individual privacy and ethical issues

Multi-omics data integration involves various sources and types of individual data, which may contain sensitive personal information, such as genetic sequences, disease status, and family history. Protecting the privacy of individual data is crucial to prevent unauthorized data access, misuse, and leakage. During the data integration process, appropriate confidentiality measures and data security mechanisms must be implemented to ensure that individual data privacy is adequately protected.

To protect individual privacy, multi-omics data integration typically requires anonymization and de-identification processes. This involves removing or replacing personal identifiers and adopting other technical measures to reduce the risk of data re-identification. However, anonymization and de-identification are not foolproof, so careful consideration and assessment of privacy risks are necessary when handling individual data.

In multi-omics data integration, researchers may need to share data to promote collaboration and enhance research effectiveness. However, data sharing must comply with legal and ethical requirements (Pang et al., 2021), including appropriate data use permissions, informed consent, and data access controls. Ensuring compliance with data sharing regulations helps protect individual privacy and maintain ethical standards.

In multi-omics data integration, the use of individual data must be based on informed consent. Researchers need to explain the purpose, risks, and benefits of data usage to individuals and obtain their explicit consent. Additionally, individuals have the right to know how their data is being used and to have appropriate control over it. Ensuring individual rights and informed consent is a crucial aspect of protecting privacy and upholding ethical standards.

5 Summary and Outlook

Multi-omics data integration holds extensive prospects for personalized therapy, but it also presents challenges that need to be addressed. By integrating various types of data, such as genomics, transcriptomics, proteomics, and metabolomics, more comprehensive and accurate individual characteristics and disease information can be obtained, aiding in precise diagnosis, personalized therapy, and drug development.

Multi-omics data integration can provide vital information and guidance for personalized therapy. Using integrated data, doctors can better understand patients' genetic backgrounds, pathological mechanisms, and disease risks, allowing for the formulation of personalized treatment plans. In drug development, multi-omics data integration can help identify new drug targets, predict drug responses, and optimize drug dosages, thereby enhancing drug efficacy and reducing adverse reactions.

However, multi-omics data integration also faces challenges. The collection and analysis of different data types involve various technical platforms and methods, requiring solutions for data consistency, standardization, and integration. Multi-omics data are typically high-dimensional and large-scale, necessitating efficient computational methods and sufficient computational resources for processing and analysis. Additionally, new algorithms and models need to be developed to accommodate the requirements of different data types and integration scenarios.

With further advancements in technology and optimization of data integration methods, the application prospects of multi-omics data integration in personalized therapy are vast. Continuous research and improvement in data integration methods and tools are needed to develop more accurate and efficient algorithms and to enhance data standardization and sharing norms. Furthermore, privacy protection and ethical issues must be adequately addressed to safeguard individual data privacy and uphold ethical principles.

Multi-omics data integration plays a crucial role in personalized therapy, providing strong support for medical decision-making and drug development. Despite facing challenges, ongoing research and innovation will likely lead to more breakthroughs in multi-omics data integration, bringing new opportunities and hope for disease prevention, diagnosis, and treatment.

Acknowledgements

I would like to express my gratitude to Ms. Liu Chuchu for her assistance throughout the stages of topic selection, data collection, and article review.

References

Afgan E., Baker D., Batut B., Van Den Beek M., Bouvier D., Čech M., Chilton J., Clements D., Coraor N., Grüning B.A., Guerler A., Hillman-Jackson J., Hiltemann S., Jalili V., Rasche H., Soranzo N.,Goecks J.,Taylor J., Nekrutenko A., and Blankenberg D., 2018, The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update, Nucleic. Acids. Res., 46(W1): W537-W544.

https://doi.org/10.1093/nar/gky379

PMid:29790989 PMCid:PMC6030816

Canzler S., Schor J., Busch W., Schubert K., Rolle-Kampczyk U.E., Seitz H., Kamp H., Bergen M.V., Buesen R., and Hackermüller, J., 2020, Prospects and challenges of multi-omics data integration in toxicology, Archives of Toxicology, 94: 371-388.

https://doi.org/10.1007/s00204-020-02656-y

PMid:32034435

Kang M., Ko E., and Mersha T.B., 2022, A roadmap for multi-omics data integration using deep learning, Briefings in Bioinformatics, 23(1): bbab454.

https://doi.org/10.1093/bib/bbab454

PMid:34791014 PMCid:PMC8769688

Li K., Du Y., Li L., and Wei, D.Q., 2020, Bioinformatics approaches for anti-cancer drug discovery, Current drug targets, 21(1): 3-17.

https://doi.org/10.2174/1389450120666190923162203

PMid:31549592

Liu J.F., Li W.L., Wang L, Li J., Li E.W., and Luo Y.M., 2022, Multi-omics technology and its applications to life sciences: a review, Shengwu Gongcheng Xuebao (Chinese Journal of Biotechnology), 38(10): 3581-3593.

Liu X., Yang Y., Ge Y.P., and Lin T., 2021, Application of artificial intelligence in clinical genomics, Zhongguo Yixue Kexueyuan Xuebao (Acta Academiae Medicinae Sinicae). 43(6): 950-955.

Manni M., Berkeley M.R., Seppey M., and Zdobnov E.M., 2021, BUSCO: assessing genomic data quality and beyond, Current Protocols, 1(12): e323.

https://doi.org/10.1002/cpz1.323

PMid:34936221

Menyhárt O., and Győrffy B., 2021, Multi-omics approaches in cancer research with applications in tumor subtype, prognosis, and diagnosis. Computational and structural biotechnology journal, 19: 949-960.

https://doi.org/10.1016/j.csbj.2021.01.009

PMid:33613862 PMCid:PMC7868685

Pang Y.J., Lyu J., Yu C.Q., Sun D.J.Y., and Li L.M., 2021, A multi-omics approach to investigate the etiology of non-communicable diseases: recent advance and applications, Zhonghua Liuxingbingxue Zazhi (Chinese Journal of Epidemiology), 42(1): 1-9.

Subramanian I., Verma S., Kumar S., Jere A., and Anamika K., 2020, Multi-omics data integration, interpretation, and its application, Bioinformatics and biology insights, 14: 1177932219899051.

https://doi.org/10.1177/1177932219899051

PMid:32076369 PMCid:PMC7003173

Wu X., Ao H.X., Gao H., and Zhu Z., 2022, Metabolite biomarker discovery for human gastric cancer using dried blood spot mass spectrometry metabolomic approach, Sci. Rep., 12(1): 14632.

https://doi.org/10.1038/s41598-022-19061-3

PMid:36030300 PMCid:PMC9420103