Introduction to RNA-Seq& Transcriptome Analysis

55
Introduction to RNA-Seq & Transcriptome Analysis Jenny Drnevich Zadeh PowerPoint by Pei-Chen Peng RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 1

Transcript of Introduction to RNA-Seq& Transcriptome Analysis

Introduction to RNA-Seq & Transcriptome Analysis

Jenny Drnevich Zadeh

PowerPoint by Pei-Chen Peng

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 1

Exercise

Use the Tuxedo Suite to:

1. Align RNA-Seq reads using TopHat (splice-aware aligner).

2. Perform reference-based transcriptome assembly with CuffLinks.

3. Obtain a new transcriptome using CuffLinks & CuffMerge.

4. Use CuffDiff to obtain a list of differentially expressed genes.

5. Report a list of significantly expressed genes.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 2

Trapnell et al., Nature Protocols, March 2012

Tuxedo Suite Bowtie and Bowtie use Burrows-Wheeler indexing for aligning reads. With bowtie2 there is no upper limit on the read length

Tophat uses either Bowtie or Bowtie2 to align reads in a splice-aware manner and aids the discovery of new splice junctions

The Cufflinks package has 4 components, the 2 major ones are listed below –

Cufflinks does reference-based transcriptome assembly

Cuffdiff does statistical analysis and identifies differentially expressed transcripts in a simple pairwise comparison, and a series of pairwise comparisons in a time-course experiment

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 3

Pipeline Overview

v

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 4

Premise

1. Procedure:

Run 1A: Allow TopHat to select splice junctions de novo and proceed through the steps

without giving the software known genes/gene models.

Run 1B: Force TopHat to use only known splice junctions (i.e. known genes/gene

models) and proceed through the steps making sure we are doing our analysis in the

context of these gene models.

2. Evaluation:

a. 2 metrics: # of mapped reads and # of significantly different identified genes

b. Compare new transcriptome to known genes.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 5

Question: Is there a difference in our results if the Tuxedo Suit is run two different ways?

sample replicate # fastq name # reads

control Replicate 1 thrombin_control.txt 10,953

experimental Replicate 1 thrombin_expt.txt 12,027

name description

chr22.fa Fasta file with the sequence of chromosome 22 from the human genome (hg19 – UCSC)

genes-chr22.gtf GTF file with gene annotation, known genes (hg19 – UCSC)

RNA-Seq: 100 bp, single end data

Genome & gene information

Input Data

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 6

Step 1A: Logging into Galaxy

Go to: http://galaxy.illinois.edu/galaxy

Click Enter

Click Login

Input your login credentials.

Click Login.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 7

Step 1B: Galaxy Start Screen

The resulting screen should look like the figure below:

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 8

Step 2A: Accessing Input Files

At the top of the page, click Shared Data.

Then click Histories.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 9

Step 2B: Accessing Input Files

Click RNA-Seq_Chr_22 Data

You should see this page.

Click Import History.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 10

Step 2C: Accessing Input Files

Click Import

You should see an imported history like the following.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 11

In this exercise, we will be aligning RNA-Seq reads to a reference genome in the absence of gene models. Splice junctions will be found de novo.

Remember, we are not going to provide any genic structure information.

.

Run 1A: de novo Alignment

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 12

Step 3A: Align Reads de novo Using TopHat

At the top right of the page, click the search box :

Type TopHat

Select TopHat under NGS: RNA Analysis

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 13

Step 3B: Align Reads de novo Using TopHat

You should a page similar to the one below. We will run TopHat first on the thrombin experimental data.

Make sure your inputs match the screenshot below:

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 14

Step 3C: Align Reads de novo Using TopHat

The rest of the page contains parameters.

We will change the following parameters:

1. Library Type: FR Unstranded

2. Minimum Intron Length: 70

3. Maximum Intron Length: 500000

4. Maximum number of alignment to be allowed: 20

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 15

Step 3C: Align Reads de novo Using TopHat

The rest of the page contains parameters.

We will change the following parameters:

5. Number of mismatches allowed in each segment alignments for reads mapped independently : 2

6. Do you want to supply your own junction data: No

7. Use coverage-based search for junctions: Yes (--coverage-search)

8. Maximum intron length that may be found during coverage search: 500000

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 16

Step 3E: Align Reads de novo Using TopHat

The rest of the page contains parameters.

We will change the following parameters:

9. Use Microexon Search: No 10. Do Fusion Search: No11. Set Bowtie2 settings: No12. Specify read group: No

Click Execute when you have set the parameters.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 17

Step 3F: Align Reads de novo Using TopHat

You will see confirmation in the Main Pane denoting which tracks have been added to run.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 18

You should see the tracks at the top of the History Pane

A gray track means the job isn't running.A yellow track means the job is running.A green track means the job is finished.

Step 3G: Align Reads de novo Using TopHatWe want to run TopHat for the control dataset now.

Navigate to the TopHat page again.

This time use 1: thrombin_control.fastq for RNA-Seq FASTQ file.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 19

Step 3H: Align Reads de novo Using TopHat2

Configure the parameters as before (below) and click execute:

1. Library Type: FR Unstranded2. Minimum Intron Length: 703. Maximum Intron Length: 5000004. Maximum number of alignment to be allowed: 205. Number of mismatches allowed in each segment alignments for

reads mapped independently : 26. Do you want to supply your own junction data: No7. Use coverage-based search for junctions: Yes (--coverage-search)8. Maximum intron length that may be found during coverage search:

5000009. Use Microexon Search: No 10. Do Fusion Search: No11. Set Bowtie2 settings: No12. Specify read group: No

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 20

Step 4A: Renaming Files

In galaxy, it is important to rename output files to something meaningful.

For example, to rename 9: Tophat_on_data4_and data2:accepted_hits

Click the pencil icon

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 21

Step 4B: Renaming Files

On the next page, enter expt_accepted_hits for the Name: field.

Click Save attributes.

Track 9 show have the name change:

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 22

Step 4C: Renaming Files

In this manner, rename the following tracks with the respective names:

5. expt_align_summary6. expt_insertions7. expt_deletions8. expt_splice_junctions

10. ctrl_align_summary11. ctrl_insertions12. ctrl_deletions13. ctrl_splice_junctions14. ctrl_accepted_hits

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 23

Step 4C: Renaming Files

Fill in required fields:

Number of comment lines: 0

Strand column (click box & select) : 5

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 24

Step 5A: Evaluating de novo Alignment

Click the eye icon 5: expt_align_summary

You should see the results on the screen, like below :

In the experimental group, 147 reads were not aligned.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 25

Step 5B: Evaluating de novo Alignment

Click the eye icon 10: ctrl_align_summary

You should see the results on the screen, like below :

In the control group, 101 reads were not aligned.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 26

In this exercise, we will be aligning RNA-Seq reads to a reference genome in the

presence of gene information. This obviates the need for TopHat to find splice

junctions de novo.

.

Run 1B: Informed Alignment

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 27

Step 6A: Informed Align Reads Using TopHat

We want to re-run the analysis for the experimental group, but using a gene-model annotation this time.

Instead of repeating the previous steps, we can save some time by clicking on the update icon on track 9: expt_accepted_hits.

Click on track 9.

Click the update icon.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 28

Step 6B: Informed Align Reads Using TopHat

Keep the same parameters as before, but change the following:

1. Do you want to supply your own junction data: Yes

2. Use Gene Annotation Model: Use a gene annotation from history

3. Gene Model Annotations:3: genes-chr22.gtf

4. Use Raw Junctions: No5. Only look for supplied junctions: No

Click Execute.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 29

Step 6C: Informed Align Reads Using TopHat

This should generate tracks 15 through 19.

Rename the tracks the following:

15. expt-genes_align_summary16. expt-genes_insertions17. expt-genes_deletions18. expt-genes_splice_junctions19. expt-genes_accepted_hits

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 30

Step 6D: Informed Align Reads Using TopHat

We want to re-run the analysis for the control

group, but using a gene-model annotation this

time.

Instead of repeating the previous steps, we can

save some time by clicking on the update icon on

track 14: ctrl_accepted_hits.

Click on track 14.

Click the update icon.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 31

Step 6E: Informed Align Reads Using TopHat

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 32

Keep the same parameters as before, but change the following:

1. Do you want to supply your own junction data: Yes

2. Use Gene Annotation Model: Use a gene annotation from history

3. Gene Model Annotations:3: genes-chr22.gtf

4. Use Raw Junctions: No5. Only look for supplied junctions: No

Click Execute.

Step 6F: Informed Align Reads Using TopHat

This should generate tracks 15 through 19.

Rename the tracks the following:

20. ctrl-genes_align_summary21. ctrl-genes_insertions22. ctrl-genes_deletions23. ctrl-genes_splice_junctions24. ctrl-genes_accepted_hits

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 33

Step 7A: Evaluating Informed Alignment

Click the eye icon 15: expt-genes_align_summary

You should see the results on the screen, like below :

In the experimental group, 39 reads were not aligned.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 34

Step 7B: Evaluating Informed Alignment

Click the eye icon 20: ctrl-genes_align_summary

You should see the results on the screen, like below :

In the control group, 27 reads were not aligned.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 35

sample # fastq name # readsUnmapped Reads

de novo Informed

experimental thrombin_expt.txt 11,679 147 39

control thrombin_control.txt 10,619 101 27

Step 8: Comparison of Alignments

There are fewer unmapped reads with the informed alignment, or Run 1B (i.e.

when we use the known genes, and known splice sites)!

TopHat’s prediction of splice junctions de novo is not working very well for this

dataset. (This is likely due to the low number of reads in our dataset.)

Conclusions

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 36

Next, we will utilize our RNA-Seq alignments to assembly gene transcripts, thereby

permitting us to get relative gene abundances between the two samples (control

and experimental).

Finding Differentially Expressed Genes

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 37

Trapnell et al., Nature Protocols, March 2012

Reminder: Cufflinks

The Cufflinks package has 4 components, the 2 major ones are listed below –

Cufflinks does reference-based transcriptome assembly

Cuffdiff does statistical analysis and identifies differentially expressed transcripts in a simple pairwise comparison, and a series of pairwise comparisons in a time-course experiment

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 38

Step 9A: Assemble Transcripts using Cufflinks

For the de-novo alignment (Run 1A) , we will run the program Cufflinks in

order to obtain gene transcripts from our aligned RNA-Seq reads .

There is no need to conduct this step for the informed alignment because

we have the locations of known genes already

Type Cufflinks into the search box.

Click on Cufflinks under NGS: RNA Analysis.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 39

Step 9B: Assemble Transcripts using Cufflinks

Choose 9: expt_accepted_hits for the BAM file.

Use the default parameters for everything except change the following:

1. Apply length correction: No Length Correction at all (use raw counts)

Ensure your parameters match up with the figure on the right.

Click Execute.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 40

Step 9C: Assemble Transcripts using Cufflinks

Go back to Cufflinks.

This time choose 14: ctrl_accepted_hits for the BAM file.

Use the default parameters for everything except change the following:

1. Apply length correction: No Length Correction at all (use raw counts)

Ensure your parameters match up with the figure on the right.

Click Execute.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 41

Step 9D: Assemble Transcripts using Cufflinks

Tracks 25 – 29 are the results of the experimental Cufflinks run.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 42

Tracks 30 – 34 are the results of the control Cufflinks run.

We will merge the assembled transcripts from the control and experimental samples next using Cuffmerge.

Step 10A: Merge Transcripts Using CuffMerge

In the search box, type Cuffmerge

Click Cuffmerge under NGS: RNA Analysis.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 43

Step 10B: Merge Transcripts Using CuffMerge

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 44

For GTF file, choose track 27: Cufflinks on data 9, which are the assembled transcripts run on the experimental accepted hits (track 9) of the de novo assembly.

Press Ctrl and choose track 32: Cufflinks on data 14, which are the assembled transcripts run on the control accepted hits (track 14) of the de novo assembly.

Choose No for the other parameters and click Execute.

Step 11A: Differential Gene Expression

For the de novo assembly, lets find out how many differentially expressed (DE) genes are present. We will use Cuffdiff to do this.

To do this, we need a GTF file and a BAM file for both the control and experimental assemblies.

We could use Cuffdiff on the informed alignments, as well, but we normally recommend using htseqcount and edgeR instead.

Type Cuffdiff into the search and click its link:

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 45

Step 11B: Differential Gene Expression

Choose track 35 for the Transcripts.

Under Condition 1:

Name: Control

Add replicate: 14: ctrl_accepted_hits

Under Condition 2:

Name: Experimental

Add replicate: 9: expt_accepted_hits

Accept the default parameters and click

Execute.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 46

Step 11C: Differential Gene Expression

When done, click the eye icon on track 45:

You should see output like the following:

Count the number of "yes" answers in the significant column (column 14) as you scroll down.

There should be 2. These are the DE genes.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 47

Conclusion

We did the following today

Use the Tuxedo Suite to:

1. Align RNA-Seq reads using TopHat (splice-aware aligner).

2. Perform reference-based transcriptome assembly with CuffLinks.

3. Obtain a new transcriptome using CuffLinks & CuffMerge.

4. Use CuffDiff to obtain a list of differentially expressed genes.

5. Report a list of significantly expressed genes.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 48

Useful linksOnline resources for RNA-Seq analysis questions –

² http://www.biostars.org/ - Biostar (Bioinformatics explained)

² http://seqanswers.com/ - SEQanswers (the next generation sequencing community)

² Most tools have a dedicated lists

Information about the various parts of the Tuxedo suite is available here -http://ccb.jhu.edu/software.shtml

Genome Browsers tutorials –

² http://www.broadinstitute.org/igv/QuickStart/ - IGV tutorials

² http://www.openhelix.com/ucsc/ - UCSC browser tutorials

(openhelix is a great place for tutorials, UIUC has a campus-wide subscription)

49

Contact us at:[email protected]

[email protected]

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018

Extra MaterialIGV

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 50

The Integrative Genomics Viewer (IGV) is a tool that supports the visualization of mapped reads to a reference genome, among other functionalities. We will use it to observe where hits were called for the de-novo alignment (Run 1A) for the two samples (control and experimental), the new transcriptome generated by CuffMerge, and the differentially expressed genes.

.

Visualization Using IGV

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 51

In this step, we will start IGV and load the chr22.fa file, the known genes file

(genes-chr22.gtf), the hits for both sample groups, and the merged transcriptome. These files

are located in [course_directory]/04_Transcriptomics/results

Step 9: Start IGV

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 52

Graphical Instruction: Load Genome

1. Within IGV, click the ‘Genomes’ tab on the menu bar.

2. Click the the ‘Load Genome from File’ option.

3. In the browser window, select chr22.fa (genome).

Graphical Instruction: Load Other Files

1. Within IGV, click the FILE tab on the menu bar.

2. Click the ‘Load from File’ option.

3. Select the genes-chr22.gtf file (known genes file).

4. Perform Steps 1-3 for the files to the right.

Files to Load

genes-chr22.fa

ctrl_accepted_hits.bam

expt_accepted_hits.bam

merged.gtf

Step 10A: Visualization With IGV

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 53

Your browser window should look similar to the picture below:

Step 10B: Visualization With IGV

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 54

Click here and type the following location of a differentially expressed gene:

chr22:19960675-19963235

Move to the left and right of the gene. What do you see?

Step 10C: Visualization with IGV

Looks like the new transcriptome (merged.gtf) compares poorly to the known

gene models. This is very likely due to the very low number of reads in our

dataset.

We can see that there are many more reads for one dataset compared to the

other. Hence, it makes sense that the gene was called as being differentially

expressed.

Note the intron spanning reads.

RNA-Seq Lab | Jenny Drnevich Zadeh | 2018 55