# What is metadata?｜中英共读稿

原作：What is metadata? — Sarah Morgan · EMBL-EBI
来源：https://www.ebi.ac.uk/training/online/courses/bioinformatics-terrified/what-makes-a-good-bioinformatics-database/describing-data-consistently/
许可：CC BY 4.0 https://creativecommons.org/licenses/by/4.0/
改编：回到本体·心晴九天，增加中文译文、学习点评与练习；保留原音画，仅调整媒体封装。不表示原作者或机构认可本站。英文来自原站字幕，中文待逐句听审。
核验日期：2026-09-16


## 00:00:00.032

So we've got a little example here. We've got someone who is ill in some way or has something wrong.

这里有一个小例子：某个人生病了，或者身体出了某种问题。

## 00:00:06.512

We've taken a sample. We've done RNA-seq. And then we have processed that and looked at the results.

我们取了样本，做了 RNA 测序，再处理数据、查看结果。

## 00:00:12.852

So we've got the raw data files generated so that's the first set of data you generated and we've got the processed files,

于是，我们有了最初生成的原始数据文件，也有了处理后的文件。

## 00:00:20.162

but we then have lots of other information to go with it. Of course, we have the actual analysis protocol.

但还需要许多配套信息，例如实际使用的分析流程。

## 00:00:24.962

So what did you do to get from the raw to the process data?

也就是，你怎样从原始数据得到处理后的数据？

## 00:00:28.982

We have the wet lab protocols which tell, you know, how you actually undertook the experiments in the lab.

还有实验室操作流程，说明实验是如何完成的。

## 00:00:34.652

So what to do experimentally? And we have the sample annotation.

包括实验中做了什么，以及样本的注释信息。

## 00:00:38.222

So, you know, who was this person? What was wrong with them? Why did we take this particular sample from them?

例如，这个人是谁、有什么问题，为什么要取这个样本？

## 00:00:45.722

So all of this gives us we've got the biological background of the actual study of doing the experimental aims.

这些信息交代研究的生物学背景，以及实验目标。

## 00:00:51.302

We've got the main experimental factors and variables we were trying to look for and in terms of submissions to an EBI database then

还有我们想研究的主要实验因素和变量。如果要向 EBI 数据库提交数据，

## 00:01:00.382

We've got things like, well, who have submitted it. So who is the person who submitted?

还需要说明是谁提交的。

## 00:01:04.672

Who's the person who's generated? Contact details?

谁生成了这些数据？如何联系他们？

## 00:01:07.572

You know, if we want to find out more about this particular dataset, you know, again, who do we speak to and are their papers?

如果想进一步了解这个数据集，该找谁？是否有相关论文？

## 00:01:16.802

So, you know, OK, I can access the raw data. I can access the process files.

我可以访问原始数据，也可以访问处理后的文件。

## 00:01:21.062

Is there a paper to go with it? So actually, I can see what their conclusion, what the researcher's conclusions and

有没有配套论文，让我了解研究人员得出了什么结论，

## 00:01:25.832

outcomes were. So all of this, you know, it does mean potentially a bit of a messy thing.

以及取得了什么结果？所有这些信息放在一起，可能显得有些杂乱。

## 00:01:34.002

Because there's so much stuff here, so much metadata that we can add to this.

因为可以添加的元数据很多。

## 00:01:39.822

And it is important because it does make it more usable by others.

但这些信息很重要，因为它们让数据更便于其他人使用。

## 00:01:45.370

But again, other people using it in future, you know, knowing exactly what you were trying to do here, who these samples were.

未来使用这些数据的人，需要准确了解你当时想做什么，以及样本的情况。

## 00:01:51.580

What did you actually do when your experiments to them get to this is all important information to

还需要了解实验具体做了什么。这些都是重要信息，

## 00:01:58.390

really get a much greater understanding of the data you've got in the data set aside for others to use.

能帮助别人更深入地理解你提供的数据集。

## 00:02:04.328

So do you think about how you structure your information?

因此，要考虑如何组织这些信息。

## 00:02:08.318

So this is actually from a data set that you will find sitting within ArrayExpress

这里是一个可以在 ArrayExpress 中找到的数据集。

## 00:02:15.558

So this is the ArrayExpress number. This is the GEO number because it came from there.

这是它的 ArrayExpress 编号；这是 GEO 编号，因为数据来自那里。

## 00:02:19.088

And there is an associated paper. This is the original sort of data.

还有一篇相关论文。这里展示的是最初提供的数据形式。

## 00:02:24.864

So all the information that came in with it. So we've got our source name.

包括随数据一起提交的全部信息，例如来源名称，

## 00:02:28.974

We've got the characteristics and then the actual descriptions.

特征以及相应的描述。

## 00:02:32.344

So all of the metadata that goes with that, which really tells us what this is, you know, what happened.

这些配套元数据交代了它是什么、发生了什么，

## 00:02:39.324

What are we looking is only one line. Now, that makes it incredibly difficult to really do anything with.

以及我们在看什么，却全都挤在一行里。这让后续使用变得很困难。

## 00:02:46.884

If you structure it slightly differently, however, you've suddenly got something where you can actually search more easily through it.

如果稍微调整结构，就能更容易地检索这些内容。

## 00:02:53.124

You could look through that data. You can always work through what we've got here.

你可以浏览这些数据，梳理其中的信息，

## 00:02:58.194

So you can then look at the particular things you want to look at.

再查看自己具体关心的部分。

## 00:03:02.004

Now things like this in terms of, you know, metadata and how this type of thing is structured.

说到元数据及其组织方式，

## 00:03:06.054

This is where standards will start to come in.

就会涉及标准。

## 00:03:08.654

the worst thing is getting to the point of submitting, then having to go backwards and try and pull out all your metadata.

最麻烦的是到了提交时，才回头搜集、整理全部元数据。

## 00:03:14.634

So start from the beginning if you know you're doing. Think about the data you need.

所以，应该从一开始就考虑需要哪些数据。

## 00:03:18.104

Think about the metadata you need. You can start to make sure you've structured it and you've got it all by collected.

也要考虑需要哪些元数据，确保有结构地收集它们。

## 00:03:22.974

The likelihood is you've got it all.

你很可能已经收集了这些信息，

## 00:03:24.274

But whether you've got it in a way that you can use is another thing. Also then you can use things like controlled vocabularies.

但它们是否处于可用的形式，是另一回事。你还可以使用受控词表。

## 00:03:32.104

So when when we are taking data submissions that are controlled vocabularies and there are also ontologies as well, which help us describe the data.

在接收数据提交时，我们会用受控词表，也会用本体来帮助描述数据。

## 00:03:40.014

So controlled vocabularies can be useful.

所以受控词表会很有帮助。

## 00:03:42.324

And that's a lovely paper which was done by Nicole Silvester two years ago from ENA, actually looking at how many different ways you can say female.

这里提到 ENA 的 Nicole Silvester 的一篇论文，研究人们有多少种方式表达 female（女性／雌性）。