A new era of bioinformatics

Full Text

SUBJECT AND TASKS OF BIOINFORMATICS, AND ITS ROLE IN MEDICINE AND FUNDAMENTAL AND APPLIED BIOLOGY

Bioinformatics is an interdisciplinary area that combines biological sciences, mathematics, statistics, and computer technology to collect, store, analyze, and interpret biological and biomedical data. This rapidly evolving field focuses on the development and implementation of algorithms and computational tools for biological data analysis, particularly in genomics, transcriptomics, proteomics, and structural and systems biology. The modern era in biology is distinguished by the rapid accumulation of vast amounts of data generated by advanced techniques such as next-generation sequencing (NGS), third-generation sequencing (TGS), structural biology, and mass spectrometry. The data generated by these techniques are often too large and complex to be managed using conventional approaches. At the same time, these massive databases may be crucial for understanding how life is organized at the molecular level. They are essential for understanding the complex biological processes that govern the structure and functioning of living systems, ranging from gene expression regulation and protein interactions to the organization of complex intracellular structures and intercellular interactions. Advances in computational techniques are critical for interpreting such complex systems and exploring the fundamental laws that control life. Research in personalized medicine, drug development, systems biology, and agricultural sciences cannot advance without the active development and implementation of bioinformatics approaches for large-scale data analysis and interpretation [1–8].

The development of bioinformatics has been largely driven by the advancement of next-generation sequencing, which is becoming increasingly accessible and integrated into routine clinical practice. Technological advances, combined with lower sequencing costs and expanded applications in a variety of fields, have significantly increased the use of these methods. In the near future, NGS and TGS technologies are expected to play a key role in shaping healthcare and become the standard for biomedical research.

In clinical diagnosis and personalized medicine, there has been substantial global increase in research based on whole genome or targeted DNA sequencing (including exomes and individual gene panels) and transcriptome sequencing. These techniques are becoming increasingly accessible to a wide range of researchers and clinical laboratories [9, 10]. NGS makes it possible to get unparalleled insights into genetic variations in human populations and investigate the mechanisms underlying hereditary diseases and cancer [11, 12]. Genome-wide association studies (GWAS) using NGS or microarray hybridization data enable identifying correlations between genetic variants and traits or diseases [13, 14]. These advancements make it possible to identify specific molecular markers of various diseases and consider their cumulative impact, allowing for personalized treatment based on patients’ individual characteristics [15]. Moreover, NGS facilitates the development of non-invasive diagnostic techniques, such as liquid biopsy, which allows monitoring disease progression and response to treatment, and non-invasive prenatal testing (NIPT) [16–18]. In addition, NGS plays a fundamental role in the development of personalized cancer immunotherapy (cancer vaccines) by identifying neoantigens expressed in tumors [19–21]. This approach not only maximizes therapeutic impact on cancer cells, but also minimizes potential side effects of broader immunotherapies that can affect healthy cells. Moreover, the development of techniques such as Chromium (10x Genomics), C1 (Fluidigm), and Seek One (Seek Gene Biotechnology) has enabled simultaneous acquisition of sequencing data for thousands of single cells (scDNA-seq and scRNA-seq) [22]. Therefore, this approach refines genetic testing to the level of single-cell analysis for assessing heterogeneity in cell populations and identifying unique events in individual cells.

Other techniques, such as ChIP-seq, ATAC-seq, and Methyl-seq and their combinations with other omics technologies, allow studying the regulation of gene expression, chromatin dynamics, and various epigenetic mechanisms [23–27]. Specialized NGS methods are being integrated into clinical practice as evidence accumulates and the clinical significance of specific mechanisms in the development of diseases is confirmed [15, 28, 29]. The molecular mechanisms of disease development are becoming much better understood at the genetic and epigenetic levels as genome sequencing and data analysis techniques advance, paving the way for a new era of precision medicine and human life extension. The new assembly T2T-CHM13, a continuous sequence of the human genome without gaps, including previously unstudied regions such as centromeres and telomeres [30, 31], has accelerated the development of genome analysis methods, including studies on functional role of repetitive sequences and the search for various structural variants. Owing to the intensive development of this field, the number of omics data collected worldwide has increased dramatically. In particular, large genomic data analysis centers generate tens to hundreds of terabytes of new data per day. By 2025, the volume of genetic data accumulated globally is expected to exceed that of information technology giants such as YouTube and X (Twitter) [32–34].

Bioinformatics is critical for understanding the structural and functional properties of proteins and peptides. Advances in mass spectrometry and other proteomic technologies enable generating complex arrays of data on protein interactions and modifications and investigating their structure [35–37]. The interpretation of these data aids in the investigation of various protein complexes and the understanding of the intricate networks of interactions between proteins, as well as proteins and nucleic acids within cells [38–42]. This information is critical for developing new drugs, studying disease mechanisms, and identifying biomarkers [43–48]. On a higher level, systems biology combines multiple layers of biological data (genomics, transcriptomics, proteomics, and metabolomics) to create complex models of biological systems. Bioinformatics tools are critical for modeling these complex systems and predicting their behavior under various scenarios.

As data accumulated, extensive multipurpose biological databases were created. These include NCBI services offering various databases and data analysis tools (https://www.ncbi.nlm.nih.gov/), UCSC genome browser, which allows visualizing genomes and contains various analysis tools (https://genome.ucsc.edu/) [49], Ensembl [50], EMBL-EBI (https://www.ebi.ac.uk/) [51], UniProt [52], Protein Data Bank [53], KEGG [54], and Enzyme Database (BRENDA) [55]. These and many other bioinformatics resources are used to annotate genomes, investigate gene function and regulation, track protein functions, metabolic pathways, and genetic interactions, and uncover new patterns by comparing biological information from various sources.

Recently, there has been a significant increase in the use of machine learning (ML) to identify patterns in complex NGS data for addressing various pharmacogenomics and oncogenetics issues [56]. Artificial intelligence (AI) technologies, a powerful tool for improving the accuracy and speed of data interpretation, are at the cutting edge of science. The integration of machine learning and AI methods facilitates the extraction of useful information, revolutionizing the analysis of omics data. It enables identifying new genetic variants relevant to disease progression, predicting disease risk, and discovering new biomarkers, which facilitates the development of personalized medicine and targeted therapeutic approaches. For example, AI-based algorithms are increasingly used in large-scale searches for new drug targets and diagnostic tools. As a result, numerous pharmaceutical companies are expanding their use of AI approaches in biomedical data processing and analysis.

ROLE OF FUNDAMENTAL MATHEMATICAL KNOWLEDGE IN THE TRAINING OF HIGHLY QUALIFIED BIOINFORMATICS SCIENTISTS

As bioinformatics becomes more important, so does the demand for qualified experts who can bridge the gap between biology and data science. The pharmaceutical and biotechnology industries are in need of professionals who can interpret genomic, transcriptomic, and proteomic data and investigate biomolecule structures for drug development and precision medicine. Bioinformaticians are also required by research institutes and laboratories to implement scientific projects and manage increasingly complex data sets. The majority of biological research today relies on bioinformatics tools to identify biomarkers, analyze next-generation sequencing findings, and model disease mechanisms.

The interdisciplinary nature of bioinformatics presents unique educational challenges, highlighting the need for specialized training programs. In addition to the fundamentals of molecular biology, bioinformatician training requires a strong mathematics background. This is due to the fact that biological data are analyzed using a variety of statistical and computational methods. The competence and correctness with which specific mathematical methods and software tools are used determines both the reliability of the analysis and the usefulness of information extracted from complex biological data. A thorough understanding of mathematics, especially statistics, probability, linear algebra, combinatorics, and graph theory, is necessary for modeling biological systems, managing large data sets, and developing predictive models.

Moreover, the complexity of biological data requires expertise in machine learning, data visualization, and programming. Programming for research purposes is typically done in languages such as Python, R, and SQL, whereas larger projects (including commercial bioinformatics software) may require C/C++, Java, or even specialized programming languages. Bioinformatics training also typically requires familiarity with cloud computing platforms for data storage and processing. All of these factors must be considered in bioinformatics education for combining theoretical and practical training in data analysis, learning advanced algorithms for interpreting omics data, and developing software. This is due to the fact that developing and optimizing new algorithms for biological data processing and analysis is one of the most important tasks of bioinformatics.

Rapid technological advancements in bioinformatics necessitate continuous education and training. This means that experts must be perpetual learners to keep up with new tools, methods, and techniques, such as deep learning approaches for analyzing omics data or advances in quantum computing. Bioinformatics education programs must evolve in tandem with these trends to ensure that graduates have up-to-date skills. The incorporation of AI and machine learning into bioinformatics is gaining traction. AI technologies have proven particularly useful in analyzing large and complex data sets in genomic analysis, single-cell transcriptome sequencing, spatial transcriptomics, and multi-omics technologies [57–61]. Deep machine learning is currently indispensable in drug development, proteomics analysis, and protein structure studies [62, 63]. DeepMind’s AlphaFold2 (and a new version, AlphaFold3, available to research institutes since November 2024) has revolutionized protein structure prediction owing to its amazing accuracy in determining 3D structures from amino acid sequences. AlphaFold2 and AlphaFold3 predict protein structure using deep learning methods, specifically neural networks trained on large data sets of known protein structures [64–68]. Demis Hassabis and John Jumper were awarded the Nobel Prize in Chemistry in 2024 for developing the AlphaFold2 algorithms, highlighting the innovative nature and significance of this technique. David Baker, who made significant contributions to the computer design of protein molecules, shared the award with them [69, 70].

Neural networks have also shown promise in designing genome editing experiments using CRISPR/Cas9 [71–73]. As these techniques gain popularity in medicine, agriculture, and ecology, bioinformaticians will play an increasingly important role in ensuring their safe and effective use.

The increasing volume and complexity of data sets inevitably raises the demand for machine learning and deep learning experts. This trend is evident in personalized medicine, where machine learning models can assist in treatment selection based on a patient’s unique genomic profile [74–79]. Multi-omics approaches that combine genomics, transcriptomics, proteomics, and metabolomics data will surely necessitate the development of new computational tools and bioinformatics methods for interpreting these complex data sets and discovering meaningful biological relationships. Bioinformatics tools are becoming more powerful and mathematically complicated, therefore training new experts requires a thorough understanding of mathematical concepts and statistical methods. Bioinformatics training programs in universities must place a greater emphasis on mathematics and statistics to provide students with the foundational skills required to traverse the field’s increasing complexity.

Importantly, the two most prominent machine learning approaches of the last 30 years, deep neural networks and support vector machines (SVM), were originally proposed and developed in the 1960s by Soviet experts in applied mathematics and mathematical statistics. Here, we will mention only the fundamental works on the first learning neural networks [80], the first deep neural networks [81], and pattern recognition [82]. Classical, strictly justified mathematical problem-solving methods remain relevant for a long time. For example, the training of modern large neural networks and large language models is based on the optimal control theory proposed in the 1950s by Pontryagin et al. [83], and the backpropagation method proposed by Galushkin [84]. Geoffrey Hinton earned the 2024 Nobel Prize in Physics for his work in training deep neural networks by applying and enhancing these methods.

CONCLUSION

In conclusion, bioinformatics is a rewarding professional path with a variety of opportunities in academic and applied research, biotechnology industry, healthcare, and entrepreneurship. The future of bioinformatics is incredibly promising, and as data accumulates and new problems emerge, bioinformaticians will continue to play a significant role in scientific progress. It is critical that the education system adapts to the changing need for highly qualified specialists in this field. The new era of bioinformatics, with the increasing use of machine and deep learning in various fields of biology and medicine, must be followed by a new era of education. We encourage non-mathematical students who want to specialize in bioinformatics to systematically improve their skills and proficiency in classical mathematical tools, particularly linear algebra, discrete mathematics, probability theory, and statistics. Universities should emphasize the mathematical component of bioinformatics by engaging relevant experts in teaching.

ADDITIONAL INFO

Author contributions: A.Yu. Aksenova, concept and design of the manuscript, collection and processing of literary data, writing the text, final editing; A.S. Zhuk — analysis of literary data, writing the text, final editing; E.I. Stepchenkova — participation in the development of the manuscript concept, final editing; V.A. Semenikhin — analysis of literary data, writing the text, intermediate editing; M.A. Langovoy — collection and processing of literary data, writing the text, final editing. The authors approved the version for publication and agreed to take responsibility for all aspects of the work, ensuring proper consideration and resolution of issues related to the accuracy and integrity of any part of it.

Acknowledgments: The authors would like to thank the staff of the RC MCT and the RC “Biobank” of St. Petersburg State University. The authors are grateful to Kirill V. Volkov for his critical comments on the manuscript.

Funding sources: This work was supported by Saint Petersburg State University, project No. 125021902561-6.

Disclosure of interests: The authors declare that there are no relationships, activities, or interests in the past three years related to third parties (commercial and non-commercial) whose interests may be affected by the content of this article.

Statement of originality: In creating this work, the authors did not use previously published information.

Generative AI: The text of this article is not the result of generative artificial intelligence.

Provenance and peer-review: This work was submitted to the journal on its own initiative and reviewed according to the standard procedure. Two external reviewers, and a member of the editorial board participated in the review.

ДОПОЛНИТЕЛЬНАЯ ИНФОРМАЦИЯ

Вклад авторов. А.Ю. Аксенова — концепция и дизайн манускрипта, сбор и обработка литературных данных, написание текста, внесение окончательной правки; А.С. Жук — анализ литературных данных, написание текста, внесение окончательной правки; Е.И. Степченкова — участие в разработке концепции рукописи, внесение окончательной правки; В.А. Семенихин — анализ литературных данных, написание текста, внесение промежуточных правок; М.А. Ланговой — сбор и обработка литературных данных, написание текста, внесение окончательной правки. Авторы одобрили версию для публикации, а также согласились нести ответственность за все аспекты работы, гарантируя надлежащее рассмотрение и решение вопросов, связанных с точностью и добросовестностью любой ее части.

Благодарности. Авторы благодарят сотрудников РЦ РМиКТ и РЦ «Биобанк» СПбГУ. Авторы благодарны Кириллу Владимировичу Волкову за критические замечания, высказанные относительно данной рукописи.

Источники финансирования. Работа выполнена при поддержке ФГБОУ ВО «Санкт-Петербургский государственный университет» (проект № 125021902561-6).

Раскрытие интересов. Авторы заявляют об отсутствии отношений, деятельности и интересов за последние три года, связанных с третьими лицами (коммерческими и некоммерческими), интересы которых могут быть затронуты содержанием статьи.

Оригинальность. При создании настоящей работы авторы не использовали ранее опубликованные сведения.

Генеративный искусственный интеллект. Текст настоящей статьи не является результатом работы генеративного искусственного интеллекта.

Рассмотрение и рецензирование. Настоящая работа подана в журнал в инициативном порядке и рассмотрена по обычной процедуре. В рецензировании участвовали два внешних рецензента и член редакционной коллегии.

About the authors

Anna Yu. Aksenova

Saint Petersburg State University

Author for correspondence.
Email: a.aksenova@spbu.ru
ORCID iD: 0000-0002-1601-1615
SPIN-code: 4914-7675

Cand. Sci. (Biology)

Russian Federation, Saint Petersburg

Anna S. Zhuk

Saint Petersburg State University; ITMO University; Vavilov Institute of General Genetics Russian Academy of Science, Saint Petersburg brunch

Email: ania.zhuk@gmail.com
ORCID iD: 0000-0001-8683-9533
SPIN-code: 2223-5306

Cand. Sci. (Biology), Assistant Professor

Russian Federation, Saint Petersburg; Saint Petersburg; Saint Petersburg

Elena I. Stepchenkova

Saint Petersburg State University; Vavilov Institute of General Genetics Russian Academy of Science, Saint Petersburg brunch

Email: stepchenkova@gmail.com
ORCID iD: 0000-0002-5854-8701
SPIN-code: 9121-7483

Cand. Sci. (Biology)

Russian Federation, Saint Petersburg; Saint Petersburg

Viacheslav A. Semenikhin

Matheomics, Skolkovo Innovation Center

Email: vasemenikhin@hse.ru
ORCID iD: 0000-0001-6923-0363
SPIN-code: 2251-5652
Russian Federation, Moscow

Mikhail А. Langovoy

Center for Artificial Intelligence SPbU

Email: mikhail@langovoy.com
ORCID iD: 0000-0002-7593-0830
SPIN-code: 6905-9451

Dr. rer. nat.

Russian Federation, Saint Petersburg

References

Supplementary files