Loading...

Big Data in Biology and Data Integration

Course video 39 of 45

The lectures in the 'Resources for Data Integration' module are about the various types of networks that are typically constructed and analyzed in systems biology and systems pharmacology. These lectures start with the idea of functional association networks (FANs). Following this lecture are several lectures that discuss how to construct FANs from various resources and how to use these networks for analyzing gene lists as well as to construct a puzzle that can be used to connect genomic data with phenotypic data.

系统生物学中的网络分析

西奈山伊坎医学院

课程 3（共 6 门，系统生物学与生物技术专项课程）

An introduction to data integration and statistical methods used in contemporary Systems Biology, Bioinformatics and Systems Pharmacology research. The course covers methods to process raw data from genome-wide mRNA expression studies (microarrays and RNA-seq) including data normalization, differential expression, clustering, enrichment analysis and network construction. The course contains practical tutorials for using tools and setting up pipelines, but it also covers the mathematics behind the methods applied within the tools. The course is mostly appropriate for beginning graduate students and advanced undergraduates majoring in fields such as biology, math, physics, chemistry, computer science, biomedical and electrical engineering. The course should be useful for researchers who encounter large datasets in their own research. The course presents software tools developed by the Ma’ayan Laboratory (http://labs.icahn.mssm.edu/maayanlab/) from the Icahn School of Medicine at Mount Sinai, but also other freely available data analysis and visualization tools. The ultimate aim of the course is to enable participants to utilize the methods presented in this course for analyzing their own data for their own projects. For those participants that do not work in the field, the course introduces the current research challenges faced in the field of computational systems biology.

查看授课大纲

教学方

Avi Ma’ayan, PhD
Director, Mount Sinai Center for Bioinformatics

脚本

So these are exciting time in the field of biology,

because we are moving from small scale biology to big data biology.

And in this lecture,

I'm going to describe to you some of the trends that are happening.

In this book, which is a business leadership book about big data.

It does not have anything to do with cell biology or biomedicine.

The revolution of big data is very well defined.

And here are some quotes from that book.

So one idea with big data is that you let the data speak.

So you enable discovery by identifying patterns without

pre-notion of what you are looking for.

Another thing that biologists are obsessed with is causality.

And this book recommends to let go of some of this obsession with causality.

So the point is that,

sometimes worrying about causality can lead us down the wrong path.

Another important aspect that biologists are already used to,

is feeling comfortable when there is some missing data.

That we can't measure everything precisely all the way to the atom level.

And the book suggests that we should feel comfortable in that area of uncertainty,

and not be so concerned with exactitude.

Finally, even though we are giving up on looking at

those specific one to one functional interactions.

We get a global perspective of the biology of the cell.

So in the next few lectures, Andrew, who is a post-doctoral fellow in our lab.

Will describe to you many of the emerging resources that can be

integrated into a global framework.

And many of those resources we processed for the enricher tool.

When in fact, those resources can be integrated into networks.

And luckily, many of them have very well structured data set.

And the IDs can be matched and

converted to standard gene IDs, drug IDs, and phenotype IDs.

And this way, you can start bringing those data

sets together to form various types of analyses.

Let's go over those seven categories.

The first category is drug and gene perturbations.

Followed by genome-wide gene expression analysis.

In many parts of the course, we discuss this type of data.

And this also is a central piece of the crowdsourcing challenges.

We also talked about Chip-Seq data.

Chip-Seq is a method to profile the binding location

of proteins on the genome.

There is also data available about cell viability after knocking down

single genes or treating cells with different drugs.

And those type of studies apply those methods to many different cell types.

So that data is a large that has on one side

drugs or gene perturbations, and on the other side different cell types.

We are also accumulating knowledge about the effect of single mutations,

or single deletion of a gene on either a human or mouse phenotype.

And those phenotype, they are relatively well organized in ontologies.

A single gene mutation or

knockout mouse, can be mapped into the resultant phenotype.

And that's an important link between a gene and the full organism phenotype.

There is also accumulating knowledge about gene expression from patients.

Specifically, complete genome Y profiling of tumors from cancer patients.

As well as gene expression profiling of post-mortem tissues from humans.

There are many databases that collect protein-protein interactions from

the literature, as well as curate, cell signaling pathways.

There are also a growing body of work

that profile protein-protein interactions using high content method.

Like immunoprecipitation followed by a mass spectrometry,

as well as these two hybrid screens.

And then finally, there are resources that describe the effects of drugs and

other toxins on the human phenotype.

And those human phenotypes are typically described as diverse events.

So all those resources can easily be converted to

either gene set libraries, which we talked a lot about in this course.

As well as network that connect genes or

other entities like drugs, or cells or phenotypes.

As well as bipartite network that connect, for example, cells and

drugs, genes and cells, genes and phenotypes, et cetera.

And then finally, all of,

all of those resources can be con, connected using those common identifiers.

And we can go and form those tripartite and multipartite relationships.

And build classifiers that can fill in the gaps and identify, for

example, side effects for drugs that haven't been marketed yet.

And this is one of the crowdsourcing projects that this course offers.

[MUSIC]

关于 Coursera

课程、专项课程和在线学位均由全世界一流大学和教育机构的顶尖授课教师教授。

Join a community of 40 million learners from around the world

Earn a skill-based course certificate to apply your knowledge

Gain confidence in your skills and further your career

Coursera 致力于普及全世界最好的教育，它与全球一流大学和机构合作提供在线课程。

© 2019 Coursera Inc. 保留所有权利。

通过 App Store 下载

通过 Google Play 获取