Scientists have long been recognized and portrayed in films as an elderly man in a lab coat perched on a bench full of bubbling fluorescent liquid. The current reality is quite different. A scientist sits in front of a monitor analyzing vast amounts of data, wearing a hoodie and increasingly becoming his data jockey. Modern labs likely consist of rows of sterile robots that manually process materials, and lab notebooks are now digitized in large data centers that hold vast amounts of information. I’m here. Today, scientific input comes from data pulled from the cloud, and like the Bunsen burner of old, algorithms drive scientific discovery.
Advances in technology, especially instrumentation, have enabled scientists to collect and process data on an unprecedented scale. As a result, scientists are now faced with massive data sets that require advanced analytical techniques and computational tools to extract meaningful insights. This also presents significant challenges. How do you store, manage, and share these large datasets to ensure high data quality and reliability?
Impact of big data on science
This growth in data is changing the way scientists conduct research, enabling new discoveries in many fields, but especially in the fields of genome and protein research. This has spurred the emergence of a whole new class of scientists who take on the role of bioinformaticians and data scientists who work with big data through the development and application of algorithms. In fact, “data scientist” has been at the top of the list of desirable jobs on career sites for the past few years. Human resources are severely lacking.
In medicine, as in other fields, not only is the volume and speed of data generation increasing, but so are the types of data collected to answer research questions. For example, flow cytometry data are fundamentally different from DNA sequencing data. Again, this is quite different from 3D models of proteins. Tools and algorithms that work for one data type are not suitable for another. Additionally, flexibility in data storage and modeling is essential for data reuse. This is especially true in predictive science, where consolidations are made between data and data types that are unrelated to the original research hypotheses.
Credit: Adobe Stock – Artificial Dreams
Look to Machine Learning and Artificial Intelligence
Technology acts like a powerful flashlight, revealing hidden patterns and insights that exist in vast amounts of data, allowing us to see and understand what was previously too dark to see. So even though the recent rise of genAI like ChatGPT has generated a lot of headlines and fueled fears about potential risks, drug discovery has been driven by artificial intelligence (AI) and machine learning (ML). It’s one setting that’s ready to make a positive impact on.
For example, during the pandemic, I had the opportunity to work with the team behind the EVE Online video game to create Project Discovery – Flow Cytometry. This is a free mini-game that has allowed tens of thousands of gamers to become citizen scientists. Using data from cell samples from patients with COVID-19 and other immune system diseases, players were trained to identify different cell patterns generated using a technique known as flow cytometry. . The game was incentivized by rewards and rankings to make it fun and challenging, but many players express their desires specifically related to participating in scientific research related to their interests and experiences. expressed his satisfaction.
To this day, players have solved millions of puzzles, representing hundreds of years of hard work. All data of the project are freely available for open science. Companies like Dotmatics can use the data to develop ML approaches to flow cytometry data analysis that are exponentially faster, cheaper, and deliver more important medical breakthroughs.
Today, both ML and AI are used in many research laboratories and universities around the world to advance discovery. The Cancer Research Center at the National Cancer Institute has developed a deep learning algorithm to improve cancer detection. For example, one model can act as a “virtual expert”.,Review MRIs for hard-to-detect cancer types, guide less experienced radiologists, and minimize error rates. Similarly, AI has been used at the University of Toronto to predict the risk of Alzheimer’s disease and at Rutgers University to predict cardiovascular disease, using advanced technology to develop cheaper and safer drugs with fewer side effects. Used by hundreds of startups to design
Big data complexity
Despite these advances, the complexity of data and the heterogeneity of the tools required to analyze these data make it difficult for researchers to effectively collaborate to generate the large datasets that AI needs. can be difficult. Initiatives such as the FAIR Guiding Principles for Scientific Data Management and Stewardship provide guidelines for improving the searchability, accessibility, interoperability, and reuse of digital assets. They are increasingly being adopted and mandated by funding agencies. Withholding of funds serves as a strong motivator in academia, but this has been found within global organizations to find large and complex datasets. It does not directly lead to pharmaceutical companies, who are perhaps even more burdened by the same underlying challenges when trying to share
The old scientific methods of using beakers and chemistry are still important, but tomorrow’s scientists will explore and understand the world around us, expanding their ambitious research into areas currently economically prohibited. But to truly harness the power of AI, invest in further improvements to the infrastructure that supports the integration, analysis, and reuse of data that is already the new frontier of scientific discovery. is needed.
About the author
Dr. Ryan BrinkmanVice President and Research Director of Dotmatics.
