Fed on Reams of Cell Data, AI Maps New Neighborhoods in the Brain

A machine learning algorithm created maps of the mouse brain with 1,300 subregions by predicting how brain cell types tend to group together, like buildings in a neighborhood.

机器学习算法通过预测脑细胞类型如何聚集在一起（例如附近的建筑物），创建了包含 1,300 个子区域的小鼠大脑地图。

Real estate agents will tell you that a home’s most important feature is “location, location, location.” It’s similar in neuroscience: “Location is everything in the brain,” said Bosiljka Tasic , a self-described “biological cartographer.” Brain injury in one spot could knock out memory; damage in another could interfere with personality.

房地产经纪人会告诉您，房屋最重要的特征是“位置、位置、位置”。神经科学领域也有类似的情况：“位置就是大脑中的一切，”自称“生物制图师”的博西利卡·塔西克 (Bosiljka Tasic) 说。脑部某一部位受伤可能会丧失记忆；他人的伤害可能会影响人格。

Neuroscientists and doctors are lost without a good map.

如果没有一张好的地图，神经科学家和医生就会迷失方向。

Researchers have been mapping the brain for more than a century.

一个多世纪以来，研究人员一直在绘制大脑图谱。

By tracing cellular patterns that are visible under a microscope, they’ve created colorful charts and models that delineate regions and have been able to associate them with functions.

通过追踪显微镜下可见的细胞模式，他们创建了描绘区域的彩色图表和模型，并能够将它们与功能联系起来。

In recent years, they’ve added vastly greater detail: They can now go cell by cell and define each one by its internal genetic activity.

近年来，他们增加了更多的细节：他们现在可以逐个细胞地进行研究，并通过其内部遗传活性来定义每个细胞。

But no matter how carefully they slice and how deeply they analyze, their maps of the brain seem incomplete, muddled, inconsistent.

但无论他们如何仔细地剖析、如何深入地分析，他们的大脑图谱似乎都是不完整、混乱和不一致的。

For example, some large brain regions have been linked to many different tasks; scientists suspect that they should be subdivided into smaller regions, each with its own job.

例如，一些大的大脑区域与许多不同的任务相关联；科学家怀疑它们应该被细分为更小的区域，每个区域都有自己的工作。

So far, mapping these cellular neighborhoods from enormous genetic datasets has been both a challenge and a chore.

到目前为止，从巨大的遗传数据集中绘制这些细胞邻域既是一项挑战，也是一件苦差事。

Recently, Tasic, a neuroscientist and genomicist at the Allen Institute for Brain Science, and her collaborators recruited artificial intelligence for the sorting and mapmaking effort.

最近，艾伦脑科学研究所的神经科学家和基因组学家塔西克和她的合作者招募了人工智能来进行分类和制图工作。

They fed genetic data from five mouse brains — 10.4 million individual cells with hundreds of genes per cell — into a custom machine learning algorithm.

他们将来自 5 个小鼠大脑（1040 万个单个细胞，每个细胞有数百个基因）的遗传数据输入到定制的机器学习算法中。

The program delivered maps that are a neuro-realtor’s dream, with known and novel subdivisions within larger brain regions.

该程序提供了神经房地产经纪人梦想的地图，在更大的大脑区域内进行了已知和新颖的细分。

Humans couldn’t delineate such borders in several lifetimes, but the algorithm did it in hours.

人类在几辈子的时间里都无法划定这样的边界，但算法在几个小时内就做到了。

The authors published their methods in Nature Communications in October.

作者于 10 月份在《自然通讯》上发表了他们的方法。

Bosiljka Tasic, a neuroscientist, genomicist, and self-described “biological cartographer,” used AI to sort through reams of cell data and build new maps of the mouse brain.

Bosiljka Tasic 是一位神经科学家、基因组学家，自称“生物制图师”，他利用人工智能对大量细胞数据进行分类，并构建了新的小鼠大脑地图。

By applying the same technique to other animals and eventually to humans, researchers hope not only to detail the brain’s finer-grained layout but also to generate and test hypotheses about how the organ’s parts operate in health and disease.

通过将相同的技术应用于其他动物并最终应用于人类，研究人员不仅希望详细了解大脑的更细粒度布局，还希望生成并测试有关该器官各部分在健康和疾病中如何运作的假设。

“We want to understand how the cells are organized in 3D space,” said Claudia Doege , a neuroscientist at Columbia University who wasn’t involved in the study.

“我们想了解细胞在 3D 空间中是如何组织的，”未参与这项研究的哥伦比亚大学神经科学家 Claudia Doege 说。

“Only if we know how they are organized can we figure out how they can potentially work with each other.”

“只有我们知道它们是如何组织的，我们才能弄清楚它们如何相互合作。”

Brain mapping is an old science, dating back to the early 1900s when the German neuroscientist Korbinian Brodmann defined regions of the cerebral cortex — the outer, thinking part of the brain.

大脑绘图是一门古老的科学，可以追溯到 1900 年代初，当时德国神经科学家 Korbinian Brodmann 定义了大脑皮层区域——大脑的外部、思考部分。

He stained human brain slices with a dye that turned genetic material violet and then studied them under the microscope, where the densities and arrangements of brain cells produced different, observable textures.

他用一种将遗传物质变成紫色的染料对人类大脑切片进行染色，然后在显微镜下研究它们，其中脑细胞的密度和排列产生了不同的、可观察到的纹理。

He traced the borders to create a map of 52 regions, known as Brodmann areas, some of which are still recognized today.

他追踪边界绘制了 52 个地区的地图，这些地区被称为布罗德曼地区，其中一些地区至今仍被认可。

I have always been fascinated and intrigued by how we can leverage AI to understand cellular organization in the brain.

我一直对如何利用人工智能来理解大脑中的细胞组织感到着迷和好奇。

Reza Abbasi-Asl, University of California, San Francisco

Reza Abbasi-Asl，加州大学旧金山分校

For decades, brain-mapping scientists wielded tools little more advanced than Brodmann’s, said Yongsoo Kim , a neuroanatomist at Penn State College of Medicine.

宾夕法尼亚州立大学医学院的神经解剖学家 Yongsoo Kim 表示，几十年来，大脑绘图科学家使用的工具并不比布罗德曼先进多少。

“What anatomists used to do is, they have a pencil, and they draw the line” between different-looking regions on brain images, he said.

他说，“解剖学家过去所做的就是，他们用一支铅笔，在大脑图像上不同的区域之间画出界限”。

One such map, the Allen Mouse Brain Common Coordinate Framework , which was published in 2020, was based on data from 1,675 mouse brains and includes more than 1,000 different areas.

其中一张地图是 2020 年发布的艾伦小鼠大脑通用坐标框架，它基于 1,675 个小鼠大脑的数据，包括 1,000 多个不同区域。

Such maps are undeniably valuable but also inevitably subjective: When Kim asked senior scientists to impart the secrets of their methods, he said the answer was often, “It’s all in my head.”

不可否认，此类地图很有价值，但也不可避免地具有主观性：当金要求资深科学家传授他们方法的秘密时，他说答案往往是：“这全都在我的脑海里。”

Recently, more advanced molecular techniques have allowed neuro-cartographers to investigate individual cells.

最近，更先进的分子技术使神经制图师能够研究单个细胞。

Under this framework, a cell’s identity is determined by which of its tens of thousands of genes are turned on, something that can be represented by the sequences of RNA molecules (copies of active DNA regions) present in the cell.

在此框架下，细胞的身份取决于其数以万计的基因中的哪一个被打开，这些基因可以通过细胞中存在的 RNA 分子（活性 DNA 区域的副本）序列来表示。

Thus, scientists can slice up a brain, measure the RNAs from each cell, and then map those genetic patterns back to the cells’ original locations.

因此，科学家可以将大脑切片，测量每个细胞的 RNA，然后将这些遗传模式映射回细胞的原始位置。

This approach has distinguished thousands of individual types of brain cells, many more than previously known.

这种方法已经区分了数千种不同类型的脑细胞，比之前已知的要多得多。

The Allen Institute’s latest mouse brain atlas, published in 2023 , includes more than 5,000 different cell types.

艾伦研究所于 2023 年出版的最新小鼠大脑图谱包含 5000 多种不同的细胞类型。

The first-draft Human Brain Cell Atlas , based on 3 million cells from the brains of three deceased people, defines 3,313 cell types.

人类脑细胞图谱初稿以三名死者大脑中的 300 万个细胞为基础，定义了 3,313 种细胞类型。

But those massive datasets didn’t yield the kind of brain cartography that Tasic sought.

但这些海量数据集并没有产生塔西克所寻求的那种大脑绘图。

The resulting maps generated regions that weren’t always “biologically meaningful,” she said.

她说，由此产​​生的地图生成的区域并不总是“具有生物学意义”。

That’s because most brain regions aren’t defined by a single cell type, and many cell types aren’t limited to one region.

这是因为大多数大脑区域并不是由单一细胞类型定义的，并且许多细胞类型并不局限于某一区域。

Instead, each area contains a mixture of cell types, including different kinds of nerve cells plus the brain’s support and immune cells.

相反，每个区域都包含多种细胞类型的混合物，包括不同种类的神经细胞以及大脑的支持细胞和免疫细胞。

“Only if we know how [cells] are organized can we figure out how they can potentially work with each other,” said Claudia Doege, a neuroscientist at Columbia University.

哥伦比亚大学神经科学家克劳迪娅·多吉 (Claudia Doege) 表示：“只有我们知道[细胞]是如何组织的，我们才能弄清楚它们如何相互合作。”

For comparison, imagine an airplane passenger looking out the window and trying to identify neighborhood boundaries within a city below.

为了进行比较，想象一下一名飞机乘客看着窗外，试图识别下面城市内的社区边界。

If the passenger focuses on just one building at a time, they can’t discern its surroundings.

如果乘客一次只关注一栋建筑，他们就无法辨别周围的环境。

To identify neighborhoods, they need to focus on how different building types group together: One neighborhood might be crowded with brownstones and playgrounds, another could be populated by mostly larger apartment buildings and bodegas, a third might be full of high-rise office complexes and restaurants.

为了识别社区，他们需要关注不同建筑类型如何组合在一起：一个社区可能挤满了褐砂石建筑和游乐场，另一个社区可能主要是较大的公寓楼和酒馆，第三个社区可能充满了高层办公楼和餐馆。

To map the brain’s subregions, Tasic needed to analyze how different cell types grouped together.

为了绘制大脑分区图，塔西克需要分析不同细胞类型如何组合在一起。

That’s not something her human brain, for all its glorious complexity, could do on its own by studying the RNA data.

尽管她的人类大脑非常复杂，但它无法通过研究 RNA 数据来完成这一任务。

Tasic needed better computational tools — and a research partner.

塔西克需要更好的计算工具和研究合作伙伴。

Tasic found the perfect collaborator in Reza Abbasi-Asl , a computational neuroscientist at the University of California, San Francisco.

Tasic 找到了完美的合作者 Reza Abbasi-Asl，他是加州大学旧金山分校的计算神经科学家。

“I have always been fascinated and intrigued by how we can leverage AI to understand cellular organization in the brain,” he said.

“我一直对如何利用人工智能来理解大脑中的细胞组织感到着迷和好奇，”他说。

To define cellular neighborhoods, Abbasi-Asl and his graduate student Alex Lee started with RNA profiles collected from 3.9 million cells in a single mouse brain.

为了定义细胞邻域，Abbasi-Asl 和他的研究生 Alex Lee 首先从单个小鼠大脑中的 390 万个细胞中收集了 RNA 谱。

They programmed a machine learning algorithm to choose one cell; its identity and gene expression would be masked.

他们编写了一种机器学习算法来选择一个细胞；它的身份和基因表达将被掩盖。

Then the AI, which they called CellTransformer, would predict that cell’s gene expression and type based on those of its neighbors, check if it had guessed right, and update its algorithm based on the result.

然后，他们称之为 CellTransformer 的人工智能会根据其邻居细胞的基因表达和类型来预测该细胞的基因表达和类型，检查它是否猜对了，并根据结果更新其算法。

By repeating this process millions of times, the algorithm learned how and where different types of brain cells group together.

通过重复这个过程数百万次，算法了解了不同类型的脑细胞如何以及在哪里聚集在一起。

From there, it could build a high-resolution map of those groups.

从那里，它可以构建这些群体的高分辨率地图。

Reza Abbasi-Asl (left) and Alex Lee at UCSF programmed a machine learning algorithm to find out how brain cell types tend to cluster together for neurobiological mapmaking.

加州大学旧金山分校的 Reza Abbasi-Asl（左）和 Alex Lee 编写了一种机器学习算法，以找出脑细胞类型如何聚集在一起以绘制神经生物学地图。

Abbasi Lab at the University of California, San Francisco

加州大学旧金山分校阿巴西实验室

Returning to that airborne city observer, what CellTransformer does is the equivalent of holding up a thumb to the window to block one building, and then predicting its type.

回到空中城市观察者，CellTransformer 所做的相当于将拇指举到窗户上挡住一座建筑物，然后预测其类型。

The surroundings provide clues as to what kind of structure fits into the neighborhood.

周围环境提供了关于什么样的结构适合该社区的线索。

Approaching brain mapping as relationships between nearby cells was the “secret sauce,” Abbasi-Asl said, that allowed the algorithm to map out meaningful neural neighborhoods, each made of a blend of different cell types.

阿巴西-阿斯尔说，将大脑映射视为附近细胞之间的关系是“秘密武器”，这使得算法能够绘制出有意义的神经邻域，每个神经邻域都由不同的细胞类型混合而成。

Depending on the level of granularity the scientists asked for, it could define anywhere from 25 to 1,300 neighborhoods in the mouse brain, though that’s not necessarily the upper limit of brain regions.

根据科学家要求的粒度级别，它可以定义小鼠大脑中 25 到 1,300 个邻域，但这不一定是大脑区域的上限。

With AI, “we see things that a human eye cannot see,” Tasic said.

塔西克说，通过人工智能，“我们可以看到人眼看不到的东西”。

Using single-cell RNA data from four additional mouse brains — including ones from male or female mice, and ones sliced from left to right or front to back — CellTransformer produced similar maps.

CellTransformer 使用来自另外四只小鼠大脑的单细胞 RNA 数据（包括雄性或雌性小鼠的大脑，以及从左到右或从前到后切片的大脑）制作了类似的地图。

This, Doege said, is excellent evidence that the technique is reliable.

多吉说，这是该技术可靠的极好证据。

While the algorithm used its predictions to group cells, it wasn’t generating wholly new maps, so it couldn’t hallucinate as some generative AI models can.

虽然该算法使用其预测对细胞进行分组，但它并没有生成全新的地图，因此它无法像某些生成人工智能模型那样产生幻觉。

Nonetheless, it was essential to compare CellTransformer’s novel output to known brain maps.

尽管如此，将 CellTransformer 的新颖输出与已知的大脑图进行比较还是很重要的。

As a trusted comparator, the team returned to the hand-drawn Allen Mouse Brain Common Coordinate Framework.

作为值得信赖的比较器，该团队回到了手绘的艾伦小鼠大脑通用坐标框架。

The CellTransformer map was a good match, laying out similar structures such as the layers in the cortex.

CellTransformer 地图非常匹配，它布置了类似的结构，例如皮层中的层。

Mouse brain maps created by the AI CellTransformer (left column) and by human scientists with the Mouse Brain Common Coordinate Framework (right column).

由 AI CellTransformer（左列）和人类科学家使用小鼠大脑通用坐标框架（右列）创建的小鼠大脑图。

Both located more than 1,000 cellular subregions in the brain (the AI found more).

两者都定位了大脑中 1000 多个细胞分区（人工智能发现了更多）。

Abbasi Lab at the University of California, San Francisco

加州大学旧金山分校阿巴西实验室

The algorithm was also able to identify new neighborhoods, regions that previous neuroscience methods, including the Allen Mouse Brain Common Coordinate Framework, had missed.

该算法还能够识别以前的神经科学方法（包括艾伦小鼠大脑通用坐标框架）所遗漏的新邻域和区域。

Take the striatum, a striped, vaguely C-shaped structure near the middle of the brain.

以纹状体为例，它是大脑中部附近的一种条纹状、隐约呈 C 形的结构。

In maps of the mouse brain, where the striatum is called the caudoputamen, “you just see one huge structure,” said Hourig Hintiryan , a neuroanatomist at the University of California, Los Angeles who wasn’t involved in the new project.

在小鼠大脑的地图中，纹状体被称为尾壳核，“你只能看到一个巨大的结构，”加州大学洛杉矶分校的神经解剖学家胡里格·辛提安（Hourig Hintiryan）说，他没有参与这个新项目。

It’s known to participate in movement, reward, and overall brain management.

众所周知，它参与运动、奖励和整体大脑管理。

How could one piece of brain perform such disparate tasks?

大脑的一块区域怎么能执行如此不同的任务呢？

CellTransformer’s explanation is that it’s not one uniform brain region after all.

CellTransformer 的解释是，它毕竟不是一个统一的大脑区域。

The map confirmed that the caudoputamen is, in fact, subdivided into smaller areas, although researchers have not yet matched each region to a function.

该地图证实尾壳核实际上被细分为更小的区域，尽管研究人员尚未将每个区域与功能相匹配。

Moreover, the new subdivisions corresponded nicely to a map that Hintiryan and colleagues published in 2016 based on an entirely different technique, which traced connections between the caudoputamen and other regions.

此外，新的细分与 Hintiryan 及其同事在 2016 年发布的地图非常吻合，该地图基于完全不同的技术，追踪了尾壳门和其他区域之间的联系。

Identifying such subregions across the brain, Hintiryan said, could resolve debates between neuroscientists who assign vastly different functions to the same large brain region.

Hintiryan 说，识别大脑中的这些子区域可以解决神经科学家之间的争论，他们为同一个大的大脑区域分配了截然不同的功能。

It seems likely that “they’re both correct, they’re just looking at different areas,” she said.

她说，“他们都是对的，只是着眼于不同的领域”。

Abbasi-Asl and Tasic were thrilled with CellTransformer’s ability to accurately match known brain cartography, and even more excited that the algorithm mapped novel subdivisions.

Abbasi-Asl 和 Tasic 对 CellTransformer 能够准确匹配已知的大脑绘图的能力感到兴奋，更兴奋的是该算法映射了新颖的细分。

For example, the brainstem’s midbrain reticular nucleus, which is involved in initiating movement, is a fairly underexplored region, Abbasi-Asl said.

阿巴西-阿斯尔说，例如，脑干的中脑网状核参与启动运动，这是一个尚未充分开发的区域。

CellTransformer picked out four new neighborhoods there.

CellTransformer 在那里挑选了四个新社区。

Each of those neighborhoods featured particularly prevalent cell types and specifically activated genes.

这些社区中的每一个都具有特别普遍的细胞类型和特定激活的基因。

They also had several cell types that earlier analyses had placed in an entirely different part of the brain.

他们还拥有几种细胞类型，早期的分析已将这些细胞类型放置在大脑完全不同的部分。

The Nature Communications paper serves mainly to introduce the CellTransformer method and show that it can find novel regions; the thousand-plus new neighborhoods still require validation.

Nature Communications论文主要介绍了CellTransformer方法并表明它可以发现新的区域；一千多个新社区仍需要验证。

As with any exploration of new territory, drawing the map is just the beginning.

与任何新领域的探索一样，绘制地图只是一个开始。

What’s most exciting is what scientists may be able to do with it.

最令人兴奋的是科学家们可以用它做些什么。

“The more granular our understanding of structure, the more specific we can get with our interrogations and interventions,” Hintiryan said.

“我们对结构的理解越细致，我们的审问和干预就越具体，”欣蒂良说。

Emerging questions center on the functions of all these neural neighborhoods.

新出现的问题集中在所有这些神经邻域的功能上。

To pinpoint what each bit does, scientists could eliminate or activate these newly identified regions in lab animals and then check for behavioral changes.

为了查明每个位点的作用，科学家可以消除或激活实验动物中这些新发现的区域，然后检查行为变化。

The real prize will be to apply CellTransformer to human brains.

真正的奖品是将 CellTransformer 应用于人类大脑。

Doege suspects that some neighborhoods will match well between mice and people, while others will diverge.

多吉怀疑，一些社区在老鼠和人类之间会很好地匹配，而另一些社区则会出现分歧。

Unfortunately, the quantity of data the algorithm needs to make accurate predictions isn’t available from human brains — at least, not yet.

不幸的是，人类大脑无法提供算法做出准确预测所需的数据量——至少目前还不行。

While the mouse brain contains about 100 million cells, the human brain has around 170 billion, and that menagerie is still undergoing genetic analysis.

小鼠大脑含有约 1 亿个细胞，而人脑则含有约 1,700 亿个细胞，而且这个细胞群仍在进行基因分析。

When sufficient amounts of that data become available, Abbasi-Asl and Tasic think CellTransformer will be up to the challenge.

当获得足够数量的数据时，Abbasi-Asl 和 Tasic 认为 CellTransformer 将能够应对挑战。

They are also interested in incorporating other technologies, such as the connection tracing used by Hintiryan, into CellTransformer.

他们还有兴趣将其他技术（例如 Hintiryan 使用的连接跟踪）整合到 CellTransformer 中。

This would be like adding streets and highways to the city neighborhoods.

这就像在城市社区中添加街道和高速公路。

And beyond the brain, the same algorithm could offer detailed cell maps of other organs, allowing scientists to compare, for example, healthy versus diabetic kidneys.

除了大脑之外，相同的算法还可以提供其他器官的详细细胞图谱，使科学家能够进行比较，例如健康肾脏与糖尿病肾脏的比较。

Human scientists simply can’t sort out these details on their own.

人类科学家根本无法独自理清这些细节。

“I see AI as kind of a helper for the human,” Kim said.

“我认为人工智能是人类的助手，”金说。

“Discovery will be accelerated in a dramatic way.”

“发现将以戏剧性的方式加速。”

Get highlights of the most important news delivered to your email inbox

获取发送到您的电子邮件收件箱的最重要新闻的摘要

Quanta Magazine moderates comments to facilitate an informed, substantive, civil conversation.

《Quanta》杂志对评论进行审核，以促进知情、实质性、文明的对话。

Abusive, profane, self-promotional, misleading, incoherent or off-topic comments will be rejected.

辱骂、亵渎、自我推销、误导、语无伦次或偏离主题的评论将被拒绝。

Moderators are staffed during regular business hours (New York time) and can only accept comments written in English.

版主在正常工作时间（纽约时间）有工作人员，并且只能接受英文评论。