New machine learning tool helps researchers demonstrate effectiveness of stem cell based models

By Emily Leclerc, Waisman Science Writer

A rosette of an organoid at 14 weeks of differentiation
A rosette of an organoid at 14 weeks of differentiation, generated by Soraya Sandoval

Studying the human brain is a particularly hard task. Studying its development is even harder. Today, many researchers are using brain organoids – miniaturized and simplified versions of organs produced in a dish typically from stem cells – as analogs for studying the development of the human brain. But is it a true comparison? A new paper by Daifeng Wang, PhD, Waisman investigator and assistant professor of biostatistics and medical informatics and computer sciences, introduces a new tool to help researchers ensure that their organoids are representative of the developing human brain. The first author of the paper is Dr. Chenfeng He, who is a postdoctoral research associate in Wang’s lab. Using machine learning, they developed an analysis technique capable of showing the similarities and differences between an organoid and a developing human brain.

Organoids have become a powerful tool to understanding human development beyond just the brain. But if the genes being expressed and the resultant changes are dissimilar from a developing brain’s gene expression programs, then the organoid is not a good model for studying human brain development. Up until now, whether or not gene expression in brains is preserved in human organoids was unclear. In this new paper, Brain and Organoid Manifold Alignment (BOMA), a machine learning framework for comparative gene expression analysis across brains and organoids, published in the journal Cell Reports Methods, Wang outlines a new computational approach for analysis of organoids and brains that showcases just how useful organoids are as models of human brain development.

A visual representation of the data comparing a human brain to an organoid after BOMA is used
A visual representation of the data comparing a human brain to an organoid after BOMA is used

The Brain and Organoid Alignment (BOMA) is a two-step process to compare gene expression data from organoids and brains. “First we use a global alignment using the existing prior knowledge about the data such as known developmental times to align individual cells between the organoid and the human brain,” Wang says. “Then we use a manifold alignment method to refine the alignment.” The first global alignment matches up cell data between the brain and the organoid but it can be messy. Afterwards, the manifold alignment creates more specific alignments and removes much of the messiness. “Unmatched cells represent unique cellular development in either the brain or the organoid,” Wang says.

For Xinyu Zhao, PhD, Waisman investigator, Jenni and Kyle Professor of Neuroscience, and collaborator on the paper, BOMA provides an important way to show that the organoids she is using in her studies are providing accurate information. “Brain organoids are in vitro models,” Zhao says. “Therefore, comparative analysis between brain organoids and developing human brains provide important information regarding the validity and limitations of organoids as a model of human brains.”

Images of human brain and organoid sections
Images of human brain and organoid sections, generated by Ryan Risgaard and Soraya Sandoval

Outside of reinforcing the notion that organoids are a legitimate route to studying human development, BOMA can have several other useful applications as well. It can be used to study organoids with manipulated genes to chart the differences in their development. BOMA can also be applied to specific patient stem cell samples to determine their potential personalized reactions and trajectories to a variety of alterations. “We can show people the whole picture of how the genes work together in the organoids versus the brain,” Wang says. BOMA is also publicly available as a web app so that other labs, scientists, and researchers can utilize the analysis technique with their own data.

The hope is that with BOMA, Waisman investigators and others, will be able to better utilize organoids in understanding human development. “Organoids are important for researchers because they allow us to study early events in neurodevelopment and, more importantly, to directly test our hypotheses by molecular manipulation,” André Sousa, PhD, Waisman investigator, assistant professor of neuroscience, and collaborator on the paper, says. “BOMA allows us to match the developmental time between in vitro cultures and the brain. This will allow researchers to design better experiments and study specific developmental features using organoids.”

The collaborative research team includes postdoctoral fellows Chenfeng He and Carissa Sirois; graduate students Noah Kalafut, Soraya Sandoval, Ryan Risgaard and Saniya Khullar; master student Chen Yang; undergraduate student Marin Suzuki;  and research scientist Xiang Huang.

Your support makes a difference. Donate now to advance knowledge about human development, developmental disabilities, and neurodegenerative diseases through research, services, training, and community outreach. DONATE NOW

Waisman Center Anniversary Logo | 1973-2023 Learn more about the Waisman Center's 50th Anniversary, including events, history, stories and images:
50 Years | 1973 - 2023