The Wonders Of AI In Genetics And Genomic Data

Artificial intelligence (AI) in its simplest definition refers to machines which have been developed to mimic human intelligence through the use of software and algorithms. AI is a broad concept encompassing analytical techniques such as machine learning (ML) and deep learning (DL). These analytical aspects unlock valuable information in various fields such as genomics which will have limitless benefits: in understanding genetic disorders; developing genetic medicines; improving cancer diagnosis and treatments; and enhancing CRISPR.

Arthur Samuel, a computer scientist, defined machine learning as “computer’s ability to learn without being explicitly programmed” [9]. ML techniques, also known as “predictive analytics or predictive modelling” [9] are initially programmed with algorithms which are used to analyse data provided and accordingly can make accurate prediction. As the availability of data increases, their performance is enhanced hence they develop ‘intelligence’ over time, having the ability to analyse complex data very quickly and accurately. ML is ideal for analyzing structured datasets such as imaging and genetics for unstructured data though such as medical journals valuable information is unlocked through the use of natural language processing (NLP) technique, another key field within AI. Converting this type of data into readable structured data enriches ML further enhancing its performance.

Save your time!
We can take care of your essay

• Proper editing and formatting
• Free revision, title page, and bibliography
• Flexible prices and money-back guarantee

Place an order

There are two main types of machine learning algorithms: supervised and unsupervised. The former is taught by examples where labelled datasets are processed and stored as models so when new data is fed their labels are predicted using the provided models. In the case of genomic sequence data, labelled DNA sequences specifying certain locations such as the start of a gene and end of a gene is required as input into the algorithm. This aids the learning of the general properties of the gene which can help identify other genes which resemble the model gene. This method is only effective when there is availability of a labelled training set. However, unsupervised ML is not provided with labelled models instead it’s left to identify patterns within datasets, and is best “If we are interested in discovering what types of labels best explain the data rather than imposing a pre-determined set of labels on the data” [10] This method is useful for the understanding of heterogeneous collection of genomic datasets and has been used in principal component analysis (PCA) to cluster genotypes from a mixture of a wide variety of genotypes in attempts to identify any relationships. Therefore, these techniques ML can be used to merely identify patterns and provide annotations for genomic sequences, and at advanced level help differentiate between different disease phenotypes and potentially identify biomarkers.

Several start-ups have taken up roles in applying these analytical techniques into practice, one example is the Canadian Deep Genomics. This company uses ML as well as DL to analyse genomic data and identify genetic causes of disease. DL (deep learning) is a subset of ML as shown in Fig 1, and “is essentially a statistical technique for classifying patterns, based on sample data, using neural networks with multiple layers” [6] unlike ML though the algorithms in DL can themselves determine whether or not a prediction is accurate. These analytical techniques are aimed at identifying patterns between diseases and mutations in large sets of genetic information, which can then be used for the selection of appropriate drug therapies, and to predict the outcomes of mutations within a cell. By understanding the relationship between the mutations and diseases, the company now aims at developing genetic medicines. It is currently focusing at using AI techniques to develop drugs aimed at treating Mendelian disorders, an inherited condition which results from a single genetic mutation and affects around 350 million individuals worldwide [13]. This revolution of using AI techniques in understanding conditions and developing drugs has engaged many due to the availability of high computing powers and powerful new algorithms coupled with the constantly increasing vast amount of data and quicker ways of sequencing genome. Brendan Frey, the CEO of Deep Genomics rightly said “the best technology we have for dealing with large amounts of data is machine learning and artificial intelligence.” [13]

Also, AI is revolutionizing the treatment of cancer. Cambridge Cancer Genomics (CCC) uses technologies such as AI to assess the effectiveness of a cancer treatment and promotes the use of personalized treatment approach. Currently the major issue in the treatment of cancer is the development of resistance as it mutates rapidly. This company, currently working on colorectal, pancreatic and lung cancer, aims to tackle the issue by using AI to predict the most appropriate treatment using cancer DNA sequencing data, and then monitor their effectiveness by analyzing tumor DNA sample obtained from liquid biopsies, so if relapse is identified the treatment plan is adjusted accordingly. This approach not only saves the patient from ineffective treatment plans, it has financial benefits for the healthcare system as well.

AI also holds benefits in CRISPR, a gene editing technology used to control the expression of genes by specifically targeting DNA sequence and deliberately inhibiting or activating them. This is promising for personalized medicine as editing individual’s genes means enhancing treatment outcomes or even protecting from toxicities. However, there are issues around the accuracy of this tool as a wrong gene could be altered resulting in harmful new mutations known as ‘off-target effects’ [1]. Tools such as Elevation use ML techniques to help predict off-target effects when editing genes which improves the accuracy and efficiency of the technology. Enhancing the accuracy of this technology means benefiting in fields such as cancer, genetic disorders and even in the discovery and development of drugs. AI also benefits this field by making the analysis of genomic data, which is constantly increasing, much efficient.

Conventional computing methods and human abilities are insufficient to accurately analyse the constantly growing amount of data, however AI enables us to qualitatively parse through such data offering limitless benefits. AI’s ability to analyse genomic data makes personalized medicine possible in many field such as cancer and genetic disorders as discussed earlier. I believe AI is and will continue to revolutionize our healthcare system as it’s a growing field, however in meantime regulations have to be implemented to ensure the security of data. Without doubt. AI is destinied to be the physician’s best assistant.

References

1. Anand (2019) Smart Data Collective Available at: https://www.smartdatacollective.com/can-ai-help-with-personalized-medicine/ (accessed 19 March 2019)
2. Bernard Marr (2018) Forbes Available at: https://www.forbes.com/sites/bernardmarr/2018/11/16/the-amazing-ways-artificial-intelligence-is-transforming-genomics-and-gene-editing/#746148b542c1 accessed 19 March 2019
3. Clara Rodriguez Fernandez (2018) LABIOTECH.eu available at: https://labiotech.eu/videos/cambridge-cancer-genomics-liquid-biopsy-ai/ accessed 19 March 2019
4. Daniel Faggella (2019) Emerj Available at: https://emerj.com/ai-glossary-terms/what-is-machine-learning/ accessed 19 March 2019
5. 5. Fei Jiang, Yong Jiang, Hui Zhi, Yi Dong, Hao Li, Sufeng Ma, Yilong Wang, Qiang Dong, Haipeng Shen, Yongjun Wang (2017), ‘Artificial intelligence in healthcare: past, present and future’ Stroke and Vascular Neurology, BMJ Journals [Online] Available at: https://svn.bmj.com/content/2/4/230 (accessed 19 March 2019)
6. Gary Marcus (2018) New York University ‘Deep learning: A Critical Appraisal’ Available at: https://arxiv.org/pdf/1801.00631.pdf (accessed 19 March 2019)
7. John Cassidy ad Harry Clifford (2018) AIMed Available at: https://ai-med.io/machine-learning-treatment-cancer/ accessed at 19 March 2019
8. Jonathan Shieber (2017) TechCrunch Available at: https://techcrunch.com/2017/08/22/cambridge-cancer-genomics-offers-a-better-way-to-gauge-effective-cancer-treatments/ accessed 19 March 2019
9. Katrina Wakefield, SAS available at: https://www.sas.com/en_gb/insights/articles/analytics/machine-learning-algorithms.html (accessed 19 March 2019)
10. Maxwell W. Libbrecht, William Stafford Noble (2015) ‘Machine learning in genetics and genomics’ PMC (PMCID: PMC5204302) [Online] Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5204302/ (accessed 19 March 2019)
11. Dr. Michael J. Garbade (2018) Medium available at: https://towardsdatascience.com/clearing-the-confusion-ai-vs-machine-learning-vs-deep-learning-differences-fce69b21d5eb (accessed 19 March 2019)
12. Samuel Mckenzie (2018) News Medical Life Sciences Available at: https://www.news-medical.net/life-sciences/Machine-Learning-in-Genetics.aspx accessed at 19 March 2019
13. Will Knight (2017) MIT Technology Review Available at: https://www.technologyreview.com/s/604305/an-ai-driven-genomics-company-is-turning-to-drugs/ (accessed 19 March 2019)