- Scientists are increasingly using artificial intelligence (AI) for predicting human behavior, and machine and deep learning can provide insight into how our health and how our body processes work.
- Machine learning also helps create predictive text models, such as those used in email and messaging services, such as WhatsApp.
- Recently, scientists investigated whether these powerful algorithms could learn protein patterns that may be responsible for neurodegenerative conditions and cancer.
- Researchers suggest this technology could lead to an increased understanding of how proteins contribute to disease beyond what the human brain can decipher.
AI is a technology that people increasingly use to gain insight into complex human processes and patterns that they could only hint at before. AI helps us make sense of huge sets of data, also known as “Big Data.”
Companies also use AI to predict consumers’ behavior. For example, this technology is how Facebook, Netflix, Amazon, and other social media and commerce websites and apps suggest what posts or products might interest users and customers.
It is also how text-based apps learn to predict a person’s next word or phrase when typing an email or text message.
Machine learning is a subset of AI. It uses large data sets to train computers to see and learn patterns in that data, analyzing and sorting accordingly, using highly complex algorithms.
These powerful algorithms have a wide range of applications, from predicting degenerative eye disease — allowing earlier and, therefore, more effective treatment to start — through to the
Recently, scientists set out to investigate whether the powerful algorithms that can predict how humans use language could also predict which proteins might be responsible for neurodegenerative conditions, such as Alzheimer’s.
They also wanted to see if a computing system they developed could identify the proteins that may contribute to cancer by observing their shape and behavior.
Their research article appears in the journal Proceedings of the National Academy of Sciences of the United States of America (PNAS).
Proteins are molecules that play a significant role in biological functions, cell structure, and tissue regulation.
When a membrane does not bind proteins in the cell, for example, in an organelle, they can form
The process that leads to their formation is called protein liquid-liquid phase separation (LLPS). Proteins in this state may influence gene expression and other behaviors within a cell in ways that we do not fully understand yet.
While some condensates are useful and allow cells to respond to changing conditions rapidly, scientists speculate that condensates may also play a role in conditions such as Alzheimer’s or cancer.
Professor Tuomas Knowles, the lead author of the study and a Fellow at St. John’s College in Cambridge, United Kingdom, explains:
“Protein condensates have recently attracted a lot of attention in the scientific world because they control key events in the cell, such as gene expression — how our DNA is converted into protein — and protein synthesis — how the cells make proteins. Any defects connected with these protein droplets can lead to diseases such as cancer.”
Because interpreting the complexities of protein behavior is critical in understanding how that process may contribute to disease, scientists wanted to see if existing AI technology used to process natural language could predict protein “language” and expression.
The team entered all data available on the known proteins so the machine could process and learn their biological language. This is pretty much how a phone app learns how to suggest words and phrases to its user.
Dr. Kadi Liis Saar, first author of the paper and a Research Fellow at St. John’s College, says, “then we were able to ask it about the specific grammar that leads only some proteins to form condensates inside cells. It is a very challenging problem, and unlocking it will help us learn the rules of the language of disease.”
As a result of their investigations, the research team developed a network called deePhase that is available to scientists worldwide. This webpage allows researchers to input a protein sequence to predict LLPS.
The research team hopes further use and development of AI, algorithms, and machine learning technology could redefine neurodegenerative disease and cancer research.
This advanced method of translating data may push past what the human brain can currently comprehend, leading to new discoveries.
Harnessing these discoveries could lead to targeted prevention and treatment options for many diseases and conditions. The goal of potential treatments would focus on correcting dysregulation inside the cells.
“Machine learning can be free of the limitations of what researchers think are the targets for scientific exploration, and it will mean new connections will be found that we have not even conceived of yet. It is really very exciting indeed.”
—Dr. Kadi Liis Saar