The Role of Computational Biology in Advancing Cancer Research

In the ever-evolving landscape of cancer research, computational biology has emerged as a vital field, leveraging the power of computational techniques to analyse and interpret the vast and complex datasets generated in biological studies. This interdisciplinary approach is revolutionising our understanding of cancer, paving the way for novel diagnostics, personalised treatments, and better patient outcomes.

Understanding Computational Biology?

At its core, computational biology involves the application of data analysis, mathematical modelling, and computer simulation to understand biological systems. In cancer research, this means utilising algorithms and models to decode genetic information, understand tumour biology, and predict responses to various therapies.

Computational Biology, Bioinformatics, and AI: What’s the Difference??

Computational Biology?

Computational biology is the broadest of the three fields, encompassing the development and application of computational techniques to solve biological problems. It includes the creation of algorithms, mathematical models, and simulations to understand complex biological systems, such as the growth and spread of cancer cells.?

Bioinformatics?

Bioinformatics is a subfield of computational biology that focuses specifically on the management and analysis of biological data, particularly genetic and genomic data. Bioinformaticians develop tools and databases to store, retrieve, and analyse DNA sequences, protein structures, and other types of biological information. In cancer research, bioinformatics is crucial for identifying genetic mutations and understanding the genetic basis of the disease.?

Artificial Intelligence?

AI, on the other hand, involves the development of systems that can perform tasks that typically require human intelligence. In the context of cancer research, AI techniques such as machine learning and deep learning are used to analyse complex datasets, recognise patterns, and make predictions. AI can assist in predicting treatment outcomes, identifying potential drug candidates, and even automating the analysis of medical images to detect cancerous tissues.?

The Importance of Computational Biology in Cancer Research

Cancer is a highly complex and heterogeneous disease. Traditional experimental methods, while invaluable, often struggle to keep pace with the complexity of the data generated. Here, computational biology steps in, providing tools to manage, analyse, and interpret these large-scale datasets. This field enables researchers to:

1. Identify Genetic Mutations: By analysing genomic data, computational biology can pinpoint specific mutations associated with different types of cancer, aiding in early diagnosis and targeted treatment strategies.
2. Understand Tumour Heterogeneity: Tumours are not uniform; they consist of various cell types with different genetic profiles. Computational methods help in characterising this heterogeneity, which is crucial for developing effective treatment plans.
3. Predict Treatment Outcomes: Machine learning models can predict how different cancers will respond to various treatments, allowing for more personalised and effective therapeutic approaches.
4. Accelerate Drug Discovery: Computational simulations can model the interactions between drugs and cancer cells, speeding up the discovery of new therapeutic agents.

Applications in Our Research Institute

At DCI, our researchers are at the forefront of computational biology, applying these advanced techniques to tackle some of the most pressing challenges in cancer research.

Personalised and Precision Medicine

Bioinformatics and computational biology research enable the use of personalised medicine approaches by identifying the best treatments for each patient or group of patients. This approach is especially successful in cases of cancers with DNA repair aberrations and exploiting synthetic lethality, such as BRCA1/2-deficient cases and PARP inhibitors.

Recent studies highlight the importance of DNA repair pathways such as Homologous Recombination (HR) and the previously understudied Nucleotide Excision Repair (NER) in a broad range of cancers. For example, a subset of gastric and oesophageal adenocarcinomas with HR deficiency responds well to platinum chemotherapy and PARP inhibitors, and NER-deficient gastric cancer cells are sensitive to cisplatin, inducing a specific type of cell death called ferroptosis.

https://www.nature.com/articles/s41698-024-00561-6

NER deficiency also makes tumour cells sensitive to irofulven, effective even in cisplatin-resistant cell lines, with minimal effects on cells with intact NER. Moreover, a novel ERCC2 mutation signature—ERCC2 being a key player in NER, whose deficiency correlates with NER deficiency—predicts response to cisplatin and irofulven.

https://aacrjournals.org/clincancerres/article/27/7/2011/671921/Identification-of-a-Synthetic-Lethal-Relationship

Detecting homologous recombination deficiency in clinical tumour samples is an important step in clinical treatment decision-making. Besides the two FDA-approved tests, there are other approaches using RAD51 foci or using copy number-based genomic scar scores and mutational signatures. These methods can involve whole exome sequencing (WES) (e.g. WES-based HRD-score) or whole genome sequencing (e.g. HRDetect) and may have different accuracy in cancer types. For example, in the case of ovarian cancer, the WES-based HRD-score has comparable accuracy to HRDetect in predicting BRCA1/2 status and long-term survival after platinum treatment.

https://aacrjournals.org/clincancerres/article/27/20/5681/671693/Comparative-Assessment-of-Diagnostic-Homologous

Deep learning, a subfield of AI, can also help detect patterns invisible to the human eye. Since 20-30% of ovarian cancer patients do not respond to platinum treatment, it is very important to identify these cases as soon as possible. At DCI, we implemented a deep learning approach to use pathological images of tumour tissue to predict response to platinum treatment in ovarian cancer and identify non-responders.

https://pubmed.ncbi.nlm.nih.gov/38883738/

These findings were made possible by the utilisation of various pipelines used in computational biology and bioinformatics, developed by the research community worldwide and in-house at DCI.

Conclusion

The integration of computational biology into cancer research is transforming our ability to understand and combat this disease. At DCI, we are proud to be at the cutting edge of this field, leveraging computational techniques to bring hope and better outcomes to cancer patients worldwide.

By harnessing the power of computational biology, we are not just keeping pace with the complexity of cancer—we are moving ahead of it, uncovering new possibilities and paving the way for breakthroughs in cancer treatment and prevention.

The next part of this two-part article will introduce the different ways Large Language Models (LLMs), the newest trend in AI, are utilised for cancer research at the Danish Cancer Institute.

Authors: Aurel?Gy?rgy Prósz, Zsófia Márta Sztupinszki and Emma Laycock