1. The effect of Next-Generation Sequencing technology on complex trait research
Challenge analysis
Ladang Auxane
Mombaerts Laurent
Uyttendaele Vincent
Presented by
December 10, 2013
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

2. TABLE OF CONTENTS Introduction Challenge analysis Applications
Optimizing parameters for study design
Storing and handling data
Mapping and aligning
Variant calling
Analyzing low frequency and rare variants
Applications
Discussion
Conclusion
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

3. INTRODUCTION What is Next-Generation Sequencing (NGS) ? Applications3
What is Next-Generation Sequencing (NGS) ?
High throughput
Low-cost
Applications
From 1970 until now
F. Sanger
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

4. CHALLENGE ANALYSIS Three mains parameters:4
1. Optimizing parameters for study design
Three mains parameters:
High cost-to-data. Only parts of the genome?
Power based on depth of coverage.
Sample selection.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

5. CHALLENGE ANALYSIS Storing and handling data Two years ago
The concept of NGS was still theoretical.
Today
Devices are operational and affordable → raw data available.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

6. CHALLENGE ANALYSIS
Production of raw data
Using a fluorescent-dye DNA sequencer
Labeling of DNA strands with 4 fluorescent dyes
Separation of fragments by electrophoresis
Monitoring by chromatography
Storage of raw data
One run can provide until 4 Tb of data
→ Requirement of a huge memory capacity
Handling of raw data
Algorithms will be applied for mapping
→ Requirement of powerful computing tools
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

7. CHALLENGE ANALYSIS De novo assembly Mapping
3. Mapping and aligning algorithms
De novo assembly
Sequencing a genome without the use of a reference genome.
Reads are assembled by an overlapping method.
Mapping
Building a sequence that is similar to a reference genome.
Reads are aligned on the backbone.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

8. CHALLENGE ANALYSIS
Improving speed and efficiency of algorithms to deal with large throughput
Detecting non-unique mapping (reads corresponding to different sequences of the reference genome)
Taking into consideration different base qualities (degrees of certainty)
Using a more accurate reference genome (including individual sequences)
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

9. CHALLENGE ANALYSIS 4. Variant calling
Distinguish true variant from sequencing or mapping errors
→ Decrease the number of false positive SNP-calls
Detecting misalignment around indels
Indel at the middle of a read : Perfect match on either side → the algorithm opens a gap.
Indel at one extremity of the read : Hard recognition of the indel → misalignment of the read → false positive SNP-call
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

10. CHALLENGE ANALYSIS
Considering different error rates depending on the base location
Nucleobases at the extremities have a higher error rate.
If misalignment : false positive confident SNP call.
SOLUTION : algorithms that consider a recalibrated “base quality score” and select only the central portion of a read.
Decreasing the number of errors introduced by PCR artefacts
PCR → not uniform cover of the reference genome → over-represented reads
SOLUTION : paired-end sequencing libraries to discard clonal reads
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

11. CHALLENGE ANALYSIS 5. Analyzing low frequency and rare variants
May be a painful step !
Single-Point
Low power
Would require hundreds of thousands of individuals
Across sample sets (composite analysis)
A bit less heavy in terms of computing time and data volume
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

12. APPLICATIONS The number of scientific publications has exploded !
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

13. DISCUSSION Development of new study design
Development of more effective methods to distinguish errors from low frequency & rare variants
Development of the most appropriate
strategy to identify one disease.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov
University of Liège

14. DISCUSSION Cost-benefit analysis
Whole genome sequencing is unlikely to be cost effective as it still presents huge challenges.
→ coupling a reduction of the costs with an increase of the efficiency and the accuracy.
→ make NGS platforms marketable, competitive and usable for clinical applications.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

15. DISCUSSION Validation analysis
Standards for NGS clinical genomics are required, for instance to validate the test accuracy.
→ important downstream impact on the patient diagnostic and management.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

16. DISCUSSION Current knowledge and research Lack of knowledge
in what a SNP implies
in how we detect interaction between genes
in which influence gene expression has
in which interpretation must be given to the genome variance …
The more we make tests, the more knowledge we get, the more associations between phenotype and genome we can do.
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

17. Enable a wide variety of applications
Conclusion
Multiple issues
Study design
Error handling
Data interpretation
Enable a wide variety of applications
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov

18. REFERENCES
A G Day-Williams, E Zeggini, The effect of Next-Generation Sequencing technology on complex trait research, Eur J Clin Invest 2011, Vol 41 :
[online on 7th December 2013]
[online on 9th December 2013]
Figures
University of Liège GBIO : Krystel Van Steen, Kyrill Bessonov