AI uses gene activity patterns to ID Parkinson’s disease pathways
Analysis helped recognize immune signaling, others as disease-related processes
An artificial intelligence (AI)-based software called NetraAI was able to separate subgroups of Parkinson’s disease patients based on their gene activity patterns, helping to better understand the complex pathways that contribute to the neurodegenerative disease.
Through the analysis, scientists identified immune signaling as an important disease-related process. The function of mitochondria, the cellular compartments that produce energy, and the microbiome, the collection of microbes living in the gut, were also implicated for some people.
The data were presented in “Using NetraAI to Discover Parkinson’s Disease Subtypes: Generative AI Reveals Transcriptomic Personas Linking Mitochondrial, Microbiome, and Immune Signaling,” at the recent International Conference on Alzheimer’s and Parkinson’s Diseases and Related Neurological Disorders (AD/PD 2024).
“The results of this analysis not only advance our understanding of PD [Parkinson’s disease] but also offer a scientific foundation for future investigations into the role of immune-related factors in neurodegenerative disorders,” Joseph Geraci, PhD, the founder and chief scientific officer at NetraMark, the software’s maker, said in a company press release. “Our unique approach opens new avenues for the diagnosis and treatment of PD, and has tremendous potential to accelerate development of urgently needed therapies for PD and other debilitating neurological conditions.”
Parkinson’s is a neurodegenerative condition that’s thought to arise from a complex interplay of genetic and environmental factors. The involvement of different biological processes likely contributes to differences in how the neurological disorder presents clinically, which complicates treatment.
AI-based algorithms, capable of analyzing and identifying patterns in large amounts of data, can help identify pathways that contribute to disease and categorize patients.
Distinguishing patterns in gene activity
NetraAI identifies subsets of patients and the biological pathways implicated in their disease based on gene activity data. Using NetraMark’s AttractorAI algorithm, it generates multiple so-called Netra-Perspectives, each one describing a group of genetic factors, or variables, that seem to distinguish patient subsets.
Each Netra-Perspective intends to “capture different aspects of complex diseases like PD,” according to the authors. By looking at multiples of them, scientists get a more holistic picture of the disease.
Every NetraPerspective includes both “explainable” subsets of patients, where particular genetic variables within the model seem to identify them, as well as “unexplainable” subsets that are not well defined by those variables. In doing so, the algorithm helps to avoid generalizations that lead to inaccurate insights or assumptions.
Here, the scientists applied NetraAI to data from 397 Parkinson’s patients and 191 healthy people that were provided by the Michael J. Fox Foundation for Parkinson’s Research.
One of the Netra-Perspectives identified increased activity of the GPATCH2L, RBBP7, and EPHA1 genes among subsets of patients. Additional analyses linked the genes to CLECB1 and IRAK3, which are involved in immune function and inflammation, and a broader network of mostly immune-associated proteins.
Another Netra-Perspective strongly implicated increased activity of the BPI gene, which is involved in protecting the body against certain types of bacteria. Other generated Netra-Perspectives implicated genes involved in the function of the microbiome and mitochondria.
The scientists theorized that BPI is increased in some patients as a protective immune response against changes in the gut microbiome that have been linked to impaired brain health.
“These findings add to the growing body of data demonstrating the power of the AttractorAI technologies on which NetraAI is based to identify the variables and provide hypotheses that explain specific patient subpopulations, even in highly complex neurological and psychiatric diseases,” Geraci said.
The technology can also be used to enrich clinical trial recruitment by identifying subsets of patients that may benefit from a therapy based on its mechanism, according to the company, which stated the NetraAI technology will also help untangle the drivers of other complex neurodegenerative diseases such as Alzheimer’s disease.
About the Author
