Math Matters at the Project ALS CORE

The key to finding better drugs at the Project ALS Therapeutics Core at Columbia (THE CORE) is developing better pre-clinical laboratory models of ALS. A range of models that actually reflect the human disease and capture the complexity of ALS are essential to bringing more rational drug candidates to people with ALS.
We’ve highlighted some of these efforts before. Drs. Hynek Wichterle and Emily Lowry, for example, are leading efforts in the in vitro unit of THE CORE to improve patient derived cellular models—so-called “ALS in a Dish”—and Dr. Alex Chavez is employing a “multiplexed system”—a space-age combination of CRISPR-Cas9 gene editing technology and genetic barcodes—to screen drugs on dozens of ALS subtypes at the same time. THE CORE’s collaboration with Drs. Lani Wu and Steven Altschuler at UCSF allows us to take another novel approach to ALS drug screening, leveraging a combination of machine learning and high-throughput single cell screening to identify potential ALS drugs.
Drs. Altschuler and Wu trained as mathematicians and worked on image recognition at Microsoft before returning to academia to apply their experience in mathematical modeling to solving complicated human diseases.  They now use a “systems biology” approach to understanding and screening potential drugs for ALS. Unlike traditional biological approaches to understanding ALS, which delve deeply into specific cellular pathways involved in the disease, the Altschuler & Wu team uses advanced imaging and computational approaches to look at the big picture of what’s happening in cells from people with ALS. At a high level, here’s how it works:

  1. Through THE CORE, the Altschuler & Wu team receives samples from dozens of people with ALS, including multiple genetic subtypes and so-called sporadic cases.
  2. The team takes high content phenotypic images of the cells—essentially, they identify hundreds of characteristics of the cells, including their shape, structure, size, and behavior, using advanced microscopy.
  3. Then, the team uses machine learning to identify what these thousands of data points generated from the images really mean. Essentially, they apply the same approach that Instagram uses to target specific ads to your feed, or that Spotify uses to recommend specific songs based on your listening habits, to identify how a cell with ALS changes under many different conditions.
  4. Using this data, the Altschuler & Wu team can learn increasingly specific information, including how cells from various ALS subtypes differ, which genetic and biological pathways are key in perturbing or rescuing cells, and exactly how potential ALS drugs impact these cells.

The Altschuler & Wu lab has successfully applied this approach to cancer, and with Project ALS support, they have now refined and optimized it for the complexities of drug discovery in ALS and other neurodegenerative diseases. Critically, THE CORE brings their systems biology approach—which provides huge amounts of information about ALS cells and how potential drugs impact them—together with advanced traditional drug screening efforts—which are slower and less scalable, but closer to the human disease—for a thorough, multidimensional evaluation of all potential ALS therapies that we evaluate.

Related Articles