Some patients with Alzheimer’s or Huntington’s diseases show symptoms early in their lives, while some others do not show symptoms until later. This difference is caused by their genetic background, and identifying these genetic modifiers provide potential drug targets whose pathways researchers hope to efficiently block. Dr. Zhandong Liu, Assistant Professor at the Center for Integrated Biomedical Research at Baylor College of Medicine, is working on drug development based on genetic modifiers that naturally occur in our body, using not only fly and mouse models but also mathematics and computer science. As genetic modifier screening is very labor-intensive and therefore expensive, Dr. Liu collaborates with top-notch neuroscientists and clinicians to build graphic models integrating multiple data types that will help advance the understanding of neurological diseases and find therapeutic targets for neurodegenerative diseases, namely Alzheimer’s and Huntington’s.

The internal collaboration between computer scientists, neuroscientists, and experimentalists within the Center facilitates effective research, encouraging feedback between researchers across disciplines inside a single, physical building. Therefore, computer scientists are able to see the experiments and understand how data is achieved, and tune their algorithms to handle the data. Here, Dr. Liu develops computational analytic tools using modern machine learning to help understand the complicated interactions of genes. Trained in both computational and biological sciences, Dr. Liu focuses on topics including genomics, transcription regulation, disease genes prioritization, and machine learning, and often examines problems in these areas using methods and models from computer science, statistics and related disciplines.

Current projects include:

• Map Building: If you want to block a disease passage, how many rows of streets do you need to block to break the disease path? Comparing genetic modifiers to “cities in a map,” Dr. Liu tries to extract the map by analyzing large amounts of data on gene expression profiles, genome sequencing of patients with neurodegenerative diseases, and clinical annotation and brain imaging of patients. He is building a computational model, or network diffusion tools, to estimate an accurate map and trying to use the map to identify therapeutic intervention points. By acquiring data of what the map looks like, enhancing the map, and analyzing the map to create solutions based on known genetic modifiers, Dr. Liu is able to extract the complicated interdependence of genes and delay the progression of diseases.
• Predicting Genetic Modifiers: Using screening results and estimating a map, Dr. Liu develops models to extrapolate the partial information of genetic modifiers to the rest of the genome. Although abundant, the data that biologists work with still only comprise of about 5% of the genome in the midst of a couple thousand candidates, and is limited in representing the entire genome data. To probe every single gene in the genome, however, is impossible. Therefore, Dr. Liu is looking for ways to accurately and efficiently investigate the genes predicted to be highly effective in modifying the disease, and extrapolating the data onto others.
• Blocking Passages: Once you know the genes you want to block, how do you use multiple drugs effectively to block a passage? Often, with complex diseases, a single compound is insufficient and multiple drugs are administered to induce desired results. However, blocking disease pathways unfortunately also block pathways for normal cells and functions. Dr. Liu therefore examines how to identify effective combinatory compounds to target these genetic modifiers with minimum off-target effects, finding the right balance.