Scientists from the Allen Institute have developed what they claim is a first-of-its-kind artificial intelligence-powered database and research tool for neuroscientists. The so-called Brain Knowledge Platform (BKP) is a comprehensive open platform containing data from over 34 million brain cells from 22 species including humans, mice, chimpanzees, and rhesus macaques. Access to the combined information from labs around the world in an accessible format could accelerate breakthroughs in a range of brain diseases.
The BKP’s developers say that the resource compiles and standardizes neuroscience data in a way that makes collaboration among labs easier. Essentially, it gives neuroscience a “common language for brain cell types much like the Human Genome Project did for genes,” said Joseph Ecker, PhD, a Howard Hughes Medical Institute Investigator from the Salk Institute for Biological Studies.
“This is special because the field has long suffered from fragmentation: different labs working in different species, with different modalities, labeling cell types differently, and with datasets that are difficult to align,” said Shoaib Mufti, MS, senior director of data and technology at the Allen Institute. Labs have historically used various methods, terminologies, and classification systems to study the brain and classify diverse cell population
The platform solves this problem by creating what its developers describe as the equivalence of a universal translator for brain science. “I am excited about the way the [BKP] will unite massive, multimodal, high-resolution datasets—including single-cell and spatial transcriptomics—all in one open, navigable environment,” Mufti said.
To develop the resource, the Allen Institute partnered with Amazon Web Services to build the core computing infrastructure that powers the BKP, and with Google to develop AI models that are designed to help scientists find patterns and connections that they might otherwise miss.
For example, a scientist studying a brain cell that seems important in Parkinson’s disease can search the platform for information about how that cell behaves in healthy brains and in people with Alzheimer’s and other neurodegenerative disorders. The AI models can compare the datasets and spot similarities and differences that could inform new therapy development. Scientists can also search for molecules and cell features using natural language queries.
The platform also provides tools that scientists can use to analyze genetic data from cells of interest in their own labs. Recent releases have added over 80 mouse whole-brain light sheet microscopy images, just over 3,000 in vivo mouse expression datasets, and over 1,700 adeno-associated viruses that represent over 1,000 cortical cell type enhancers.
“We’ve created and shared high-quality brain maps since our first mouse brain atlas in 2007,” says Hongkui Zeng, PhD, executive vice president and director of Brain Science at the Allen Institute. “The Brain Knowledge Platform enhances those maps with the novel understanding of cell types we’ve been developing with others in the NIH’s BRAIN Initiative. Like topography on Google Maps, cell type information adds multiple new layers to our maps, helping scientists design better experiments and glean new insights.”
Importantly, the platform connects basic brain research data to medical treatments. Because the platform includes data from both healthy and diseased brains, scientists can identify which cells are impacted in disorders such as Alzheimer’s and Parkinson’s and then test potential treatments on those cells.
The post Brain Knowledge Platform Provides AI Models, Data for Disease Studies appeared first on GEN – Genetic Engineering and Biotechnology News.
