1. Maximize read usage through mapping strategies

2. Key concepts of session
The settings you use MATTER.
More reads mapping is not ALWAYS better.
Sometimes you need to understand what you threw away.
Mapping is yet another quality control step.

3. What are we doing with mapping?
How do we align the bag of reads to the reference?
Efficiently (memory and time)
Account for inexact matches and ambiguity?
Traditional sequence alignment (BLAT, Blast) are too slow for millions of short reads.

4. Short read mapping Input: Output:
A reference genome
A collection of many bp tags/reads
User-specified parameters
Output:
One or more genomic coordinates for each tag
In practice, not all reads successfully map to the reference genome. Why?

5. What makes it hard?
Inexact matches
Multiple mapping ?

6. Indexing
The key to speeding up matching short reads is to tightly INDEX the genome.
There are multiple strategies for building indicies:
> 2000X faster than BLAST
Suffix Tree:
Suffix Array:
Hashing:

7. Bowtie
Indexes the reference genome using a scheme based on the Burrows-Wheeler transform (BWT) which is very space efficient.
A quality-aware backtracking algorithm that allows mismatches and favors high-quality alignments.
Double indexing', a strategy to avoid excessive backtracking

8. Bowtie caveat
“If one or more exact matches exist for a read, then Bowtie is guaranteed to report one, but if the best match is an inexact one then Bowtie is not guaranteed in all cases to find the highest quality alignment.”
…unless you use the MUCH slower “best” option

9. Mature RNA presents a unique problem for read mapping …

10. Read mapping
exon mapping
exon-exon junction
Unlike DNA-Seq, when mapping RNA-Seq reads back to reference genome, we need to pay attention to exon-exon junction reads

11. Three mapping strategies.
De novo – can align assembled products to genome if have reference
Transcriptome – rely on ANNOTATION or do isoform inference and then rely on inferred isoforms.
Reference genome – this is TOPHAT/CUFFLINKS sort of approach; Data driven.
Diagrams from: Cloonan & Grimmond, Nature Methods 2010

12. Many splice junctions per gene

13. Mapping software …

14. Alignment quality score
Base quality values and mismatch positions in a candidate alignment are
used to assign a probability value
Probability reflect likelihood that candidate position in genome would
give rise to the observed read if its bases were sequenced at error rates
corresponding to the read’s quality values. Should also reflect “uniqueness”.
Alignment score for a read is computed from probability values of all candidate alignments.
If there are two candidate alignments for a read with probabilities
values 0.9 and 0.3:
0.9/( ) = 0.75, chance highest scoring alignment is correct
, chance highest scoring alignment is wrong
Alignment score = -10 log(0.25) = 6.

15. Sequence Alignment/Map (SAM)

16. Sequence Alignment/Map (SAM) format
Standard format for reporting short read alignment data
BAM is compressed version
Header
Alignment
info
Header:

17. Read Information

18. Read Information

19. Mapping Flags

20. How is mapping a quality control step?
Poor quality reads will not map with the correct settings
If too many reads are thrown away, it may may indicate a problem in pre-mapping qc (protocol, trimming)
RED = not mapped
YELLOW = mapped, duplicated
BLUE = uniquely mapped

21. Mapping : from fastq to tdf
After successful trimming, it’s time to map
Again, there are a number of different mapping tools all with pros and cons
We will use Bowtie2

22. In part one of this script, we will:
Set our Bowtie2 parameters
Setting mapping sensitivity will largely affect time it takes to run the file (fast, sensitive, very sensitive)
Generate mapping stats

23. In the next part of this script, we will:
Convert .sam to .bam (binary sam, quicker processing) and index  see handout
Generate bedGraph file which gives information about read coverage

24. Lastly, we will:
Read count correct – adjust for read depth post-mapping using flagstat file
Convert .bedGraph to .tdf for rapid loading into IGV
To run (after adjusting rootname/project – refdir is the same for everyone):
$ bash mapping.sh

25. Open tdfs in IGV Path = $ cd data/nascent-ws/Mapped/tdfs/ Open all SRRs with .tri.tdf : see dowell.colorad.edu/HackCon/pages/visualization.html for IGV tips and tricks

26. Evaluating your mapping
Poor quality reads will not map
Too many duplicates may indicate a problem with the library or sequencing
better to run a small “test” sample (~10M reads) if you have time and $$
Look at your sample!!

27. Post-mapping : additional QC
RSeQC – assessing where reads are mapping in the genome
SCRIPT: rseqc.sh
Preseq – determining sample complexity (how many unique reads)
SCRIPT: preseq.sh