Summary: A recent study demonstrates how deep learning algorithms based on artificial intelligence ( AI ) algorithms can identify the regulatory mechanisms that affect different species of brain cells. By analyzing people, rat, and meat brains, researchers found that some head cell types remain very conserved over 320 million years, while others have evolved exclusively.
This regulation script provides new tools for studying protein rules in health and disease as well as providing new insights into brain evolution. The findings demonstrate how artificial intelligence can identify both conserved and different genetic instructions that govern brain function across different species.
The study also has potential for improving our understanding of neurological disorders by fusing genetic variants with mental characteristics. To examine the brains of different animals and people with disabilities like Parkinson’s, experts are now expanding their versions.
Major Information
- Biological Conservation: Some head body regulatory rules have remained unchanged for over 320 million times.
- AI-Powered Discovery: Deep learning models help convey biological rules across species.
- Disease Software: These designs could be used to analyze neurological problems like Parkinson’s.
Origin: VIB
A French research team examines how biological switches controlling protein action determine brain cell types across species in a new study published in Science.
They conducted deep learning models using data from people, rat, and meat brains, and found that while some mobile types are remarkably conserved between birds and mammals over millions of years, others have evolved in a different way.
The findings provide important tools for understanding how gene rules influences various cell types, species, or condition states, as well as new insights into brain evolution.
Our head, and by extension our whole body, is made up of many different types of cells. Despite having the exact DNA, each of these body types has a unique shape and function.
Scientists have been attempting to solve a difficult issue that has been gathered from small DNA segments, which control which genes are turned on or off, for years.
Each type of brain cell uses just the right biological guidance from the genome to fulfill its distinctive part due to the fine-tuned regulation of these switches. The distinctive designs that these biological switches exhibit serve as a regulatory code, according to researchers.
Artificial to crack the code
Prof. Stein Aerts and his team at VIB. The basic rules of this regulation code are studied by AI and the VIB-KU Leuven Center for Brain & Disease Research, and how it might affect conditions like cancer and mental problems.
They develop deep learning techniques to help understand the enormous amount of data on protein restriction that thousands and thousands of unique cells are able to obtain from them.
” Deep-learning models that use the DNA sequence code have greatly improved our ability to identify regulatory mechanisms across various cell types,” explains Aerts.
We wanted to know if this regulation password could provide any additional information on how these body types are maintained across species.
In the mind, for instance, is where a question like that is very important. Despite shared evolutionary paths, the brains of animals and birds display a strikingly different physiology.
Aerts and his team had now used deep learning models to determine whether the existing differences and similarities are reflected in different or shared regulatory standards.
Tool to research development
In the Aerts test, Nikolai Hecker and Niklas Kempynck, both doctorate and PhD students, created and implemented machine learning models to evaluate and identify various cell types from people, rat, and poultry brains over a period of roughly 320 million years.
But before they could really evaluate, they first had to understand the chicken brain’s cell type composition, but they created a detailed genomic atlas.
” Our research demonstrates how we can use heavy learning to identify and evaluate different cell types based on their regulatory codes,” Hecker says. ” We can use these rules to assess the genomes of various species, discover which regulatory codes have survived over time, and discover how body types have evolved.”
The study found that while some regulatory body types are closely related between birds and mammals, others have evolved separately. Importantly, the regulatory codes for selected parrot neurons resemble those of deep-layer cells in the vertebrate neocortex.
” Looking straight at the governmental code presents a major edge”, adds Kempynck,” It can tell us which governmental guidelines are shared across varieties, even if the DNA sequence itself has changed”.
Tool to study disease
Beyond the study of evolution, this regulatory information is useful. Aerts and his team have already established that mammals and zebrafish share the same regulatory codes for melanoma (skin cancer ) cell states as their counterparts. Additionally, they discovered variants in melanoma patients ‘ genomes.
The brain cell type models used in the current study provide useful tools for studying the impact of genomic variants and their associations with mental or cognitive traits and disorders.
Aerts asserts that “models that learn the genomic regulatory code have the potential to screen genomes and assess whether or not particular cell types or states exist in any species.” This would be a useful resource for studying and better understanding disease.
To the zoo
Aerts and his team are already putting their models to the test, he says:” Both fronts are being used.”
” We are now expanding our evolutionary modeling to many more animal brains, including different types of fish to dear, hedgehogs, and capibaras,” in collaboration with Zoo Science and Wildlife Rescue Center. At the same time, we’re also exploring how these AI models can help to unravel genetic variation linked to Parkinson’s disease”.
About this genetics, AI, and evolutionary neuroscience research news
Author: Gunnar De Winter
Source: VIB
Contact: Gunnar De Winter – VIB
Image: The image is credited to Neuroscience News
Original Research: Closed access.
The study by Stein Aerts and colleagues,” Enhancer-driven cell type comparison reveals similarities between bird pallium and mammalian cells.” Science
Abstract
Comparing the mammalian and bird pallium cell types using enhancer-driven cell types reveals similarities.
INTRODUCTION
Gene regulatory networks, which are composed of cell type-specific transcription factor ( TF ) binding to genomic enhancer regions, determine the identity of cell types.
The TF binding site configurations create cell type-specific enhancer codes. The ability to model and characterize enhancer codes at nucleotide resolution is provided by deep learning models trained on single-cell data. The mammalian telencephalon, which includes the pallium and makes up a significant portion of the forebrain, has not yet been characterized for enhancer codes at this resolution.
There are significant neuroanatomical differences between mammals and nonmammalian vertebrates in the pallium. Most noticeable, the mammalian pallium contains a six-layered neocortex that is absent in all nonmammalian vertebrates, such as birds.
A decade-long debate surrounds the morphologies of the mammalian and bird pallium. Whether enhancer codes are present in all vertebrate brains and whether they provide information on how to determine homology between species at the cell level are still unknown.
RATIONALE
We created single-cell multiome (scMultiome ) and spatially resolved transcriptomics data for the chicken telencephalon to characterize and compare the enhancer codes of brain cells found in birds and mammals.
As a baseline to map cell type similarities between species, we compared the transcriptomes of telencephalon cell types between human, mouse, and chicken. Then, to determine potential genomic enhancer regions and to assess their cell type specificity, we used variable chromatin accessibility as a proxy.
Next, we trained sequence-based deep learning models on these regions to infer cell type–specific enhancer codes for the human, mouse, and chicken telencephalon. To compare cell types among species, we created three metrics that make use of enhancer codes.
RESULTS
Excitatory neurons of the chicken telencephalon distinctly localize to pallial neuroanatomical regions, including the mesopallium, entopallium, hyperpallium, and nidopallium.
Nonneuronal and -aminobutyric acid-mediated ( GABAergic ) cell types exhibit a high degree of similarity across birds and mammals, which is demonstrated by conserved TF combinations for these cell types.
On the other hand, the enhancer codes of avian and mammalian excitatory neurons exhibit greater divergence. Based on developmental trajectories and brain circuitry, these matches only partially match current evolutionary models for homologies between vertebrate pallial cell types.
We found that the mammalian deep-layer excitatory neurons are most similar to mesopallial neurons, and mammalian neocortical upper-layer, piriform cortex, and amygdalar neurons are most similar to hyper- and nidopallial neurons.
We conducted in vivo enhancer reporter assays to validate the predicted correlations between mammalian and bird cell types. When examined in mouse brains, we demonstrate that chicken enhancer sequences exhibit activity in the corresponding mammalian telencephalic cell types.
CONCLUSION
Our study demonstrates that enhancer codes can be used to infer cell type correspondences between species that are compatible with transcriptomic comparisons.
In contrast to the mammalian and avian telencephalon, joint analyses of transcriptomes and deep learning-based enhancer codes reveal both expected and unanticipated correlations between cell types, indicating conserved regulatory programs that were likely initiated by common amniote ancestors and have since been re-opted or diversified.
The genomic regulatory code-based approaches that have been proposed generally apply and can be used to characterize and compare cell types across species.
