Objectives
In this lab, you will explore a popular fast mapper called BWA. Simulated RNA-seq data will be provided to you; the data contains 75 bp paired-end reads that have been generated in silico to replicate real gene count data from Drosophila. The data simulates two biological groups with three biological replicates per group (6 samples total). The objectives of this lab is mainly to:
- Learn how BWA works and how to use it.
Introduction
BWA (the Burrows-Wheeler Aligner) is a fast short read aligner. It's the successor to another aligner you might have used or heard of called MAQ (Mapping and Assembly with Quality). As the name suggests, it uses the burrows-wheeler transform to perform alignment in a time and memory efficient manner.
BWA Variants
BWA has three different algorithms:
- For reads upto 100 bp long:
- BWA-backtrack : BWA aln/samse/sampe
- BWA-backtrack : BWA aln/samse/sampe
- For reads upto 1 Mbp long:
- BWA-SW
- BWA-MEM : Newer! Typically faster and more accurate.
Get your data
Six raw data files have been provided for all our further RNA-seq analysis:
- c1_r1, c1_r2, c1_r3 from the first biological condition
- c2_r1, c2_r2, and c2_r3 from the second biological condition
| Code Block | ||
|---|---|---|
| ||
cds
cd my_rnaseq_course
cp -r /corral-repl/utexas/BioITeam/rnaseq_course_2015/bwa_exercise . &
cd bwa_exercise |
Run BWA
Load the module:
| Code Block |
|---|
module load bwa/0.7.7 |
There are multiple versions of BWA on TACC, so you might want to check which one you have loaded for when you write up your awesome publication that was made possible by your analysis of next-gen sequencing data.
| Expand | ||
|---|---|---|
| ||
|
...
You can see the different commands available under the bwa package from the command line help:
| Code Block |
|---|
...
bwa
|
Try to figure out how to index and map from the command line help:
| Code Block |
|---|
bwa
|
...
Part 1. Create a index of your reference
NO NEED TO RUN THIS NOW- YOUR INDEX HAS ALREADY BEEN BUILT!
| Code Block |
|---|
bwa index -a bwtsw reference/genome.fa
|
Part 2a. Align the samples to reference using bwa aln/samse/sampe
We will be using this set of commands (with options that you should try to figure out) in this order, on each sample:
...
bwa aln bwa samse or sampe
...
What's going on at each step?
Remember to use the option that enables multithreading, if there is one, for each BWA command.
First, run the index command (index) on the reference file. This is fast, so you can run it interactively.
BWA doesn't give you a choice of where to create your index files. It creates them in the same directory as the FASTA that you input. So copy the FASTA in your intro_to_mapping directory to your new bwa directory:
| Expand | ||
|---|---|---|
| ||
|
Then, run the index command using the copied FASTA as input.
| Expand | ||
|---|---|---|
| ||
|
Take a look at your output directory using ls bwa to see what new files appear after indexing.
...
Let's submit the bwa aln job
| Warning | |||||||
|---|---|---|---|---|---|---|---|
| |||||||
Create a
|
*.sai file is a file containing "alignment seeds" in a file format specific to BWA. We still need to extend these seed matches into alignments of entire reads, choose the best matches, and convert the output to SAM format. Do we use sampe or samse?
Lets submit the bwa sampe job, but have it be on hold till previous job is finished.
| Warning | |||||
|---|---|---|---|---|---|
| |||||
Create a
|
...
|
...
Why did we use -t 6 instead of -t 12 for multithreading? Both of our commands are going to go to a single node on Lonestar, so they should share the 12 available cores.
Again, take a look at your output directory using ls bwa to see what new files have appeared. What is a *.sai file? It's a file containing "alignment seeds" in a file format specific to BWA. Many programs produce this kind of "intermediate" file in their own format and then at the end have tools for converting things to a "community" format shared by many downstream programs.
We still need to extend these seed matches into alignments of entire reads, choose the best matches, and convert the output to SAM format.
...
|
Part 2b. Align the samples to reference using bwa mem
| Warning | |||||||
|---|---|---|---|---|---|---|---|
| |||||||
Create a
|
|
Since these will take a while to run, you can look at already generated results at: /corral-repl/utexas/BioITeam/rnaseq_course_2015/bwa_exercise/results/bwa
Help! I have a lots of reads and a large number of reads. Make BWA go faster!
Use threading option in the bwa command ( bwa -t <number of threads>)
- Split one data file into smaller chunks and run multiple instances of bwa. Finally concatenate the output.
- WAIT! We have a pipeline for that!
- Look for runBWA.sh in $BI/bin (it should be in your path)
Now that we are done mapping, lets look at how to assess mapping results.