Transcript Lecture_6 (2014)x

High Throughput Sequencing
Agenda
•
•
•
•
Introduction to sequencing
Applications
Bioinformatics analysis pipelines
What should you ask yourself before
planning the experiment
Introduction to sequencing
What is sequencing?
Finding the sequence of
a DNA/ RNA molecule
What can we sequence?
http://cancergenome.nih.gov/newsevents/multimedialibrary/images/CancerBiology
Sanger sequencing
• Up to 1,000 bases molecule
• One molecule at a time
• Widely used from 1970-2000
• First human genome draft was
based on Sanger sequencing
• Still in use for single molecules
http://www.genomebc.ca/education/articles/sequencing/
High Throughput Sequencing
Next Generation Sequencing (NGS) / Massively parallel sequencing
• Sequencing millions of molecules in parallel
• Do not need prior knowledge of what you’re
sequencing
Platform
Read length
454 sequencing Up to 1,000 bp
SOLiD
50-75 bp
HiSeq
100-150 bp
No. of reads
per run
~1 M (1 million)
~1 G (1 billion)
~0.5 G
We will discuss Illumina’s platform only
Sequencing Workflow
1.
2.
3.
4.
5.
Extract tissue cells
Extract DNA/RNA from cells
Sample preparation for sequencing
Sequencing
Bioinformatics analysis
Why is it important to understand
the “wet lab” part?
Sample Prep
Random shearing of
the DNA
Size selection
Amplification
Adding adaptors
and barcodes
Sequencing
Sequencing process
Sequencing process
Leave only sequences
from one direction
Sequencing process
Sequencing process
Sequencing process
Applications
DNA sequencing
• Resequencing – sequencing the genome of an
organism with a known genome
• Exome sequencing / Targeted sequencing –
sequencing only selected regions from the
genome
• De-novo sequencing– sequencing the genome of
an organism with a unknown genome
RNA-Seq
Sequencing of mRNA extracted form the cell to
get an estimate of expression levels of genes.
Counting vs. Reading
RNA-Seq vs. DNA-sequencing
ChIP-Seq
Sequencing the regions in the genome to which
a protein binds to.
Basic concepts
Insert – the DNA fragment that is used for sequencing.
Read – the part of the insert that is sequenced.
Single Read (SR) – a sequencing procedure by which
the insert is sequenced from one end only.
Paired End (PE) – a sequencing procedure by which the
insert is sequenced from both ends.
Bioinformatics Analysis Pipelines
Demultiplexing
Unknown inserts
Lane
Demultiplexing
Mapping
Sample
Reference
Genome
Example of mapping parameters:
• Number of mismatches per read
• Scores for mismatch or gaps
Mapping parameters affect the rest of the analysis
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
Reference
Genome
?
Reference
Genome
𝒂𝒗𝒆𝒓𝒂𝒈𝒆 𝒄𝒐𝒗𝒆𝒓𝒂𝒈𝒆 =
𝑟𝑒𝑎𝑑 𝑙𝑒𝑛𝑔𝑡ℎ ⋅ 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑟𝑒𝑎𝑑𝑠 ⋅ % 𝑢𝑛𝑖𝑞𝑢𝑒𝑙𝑦 𝑚𝑎𝑝𝑝𝑒𝑑 𝑟𝑒𝑎𝑑𝑠
𝑔𝑒𝑛𝑜𝑚𝑒 𝑠𝑖𝑧𝑒
Resequencing/ Exome Pipeline
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
Coverage profile and
variant calling
G
…ACTTCGTCGAAAGG…
Reference
Genome
Demultiplexing
Frequency >= 20%
Mapping
Removing duplicates and
non-unique mappings
Coverage profile and
variant calling
…ACTTCGTCGAAAGG…
Variant filtering
Reference
Genome
Coverage >= 5
…ACTTCGTCGAAAGG…
Reference
Genome
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
G
Variant calling
A
Variant filtering
Genes and known variants
…ACTTCGTCGAAATG…
Gene X
…GTCCCGTGATACTCCGT…
rs230985
Reference
Genome
Resequencing results
Example for further analysis
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
Coverage profile and
variant calling
Variant filtering
Genes and known variants
Finding suspicious
variants
Recessive disease:
1. Variant not in known databases
2. Homozygous variant shared by all
affected individuals
3. Same variant appears in healthy parents
at heterozygous state
4. Healthy brothers can be heterozygous
to the same variant
Dominant disease:
1. Variant not in known databases
2. Heterozygous variant shared by all
affected individuals
3. The variant doesn’t appear in healthy
individuals
Quality control steps in the pipeline
Demultiplexing
QC
Mapping
QC
Removing duplicates and
non-unique mappings
Coverage profile and
variant calling
QC
QC
Variant filtering
QC
Genes and known variants
Finding suspicious
variants
How is de-novo assembly different
from resequencing analysis ?
RNA-Seq Pipeline
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
FPKM Normalization:
𝑟𝑎𝑤 𝑐𝑜𝑢𝑛𝑡
𝑔𝑒𝑛𝑒 𝑙𝑒𝑛𝑔𝑡ℎ ⋅ 𝑠𝑎𝑚𝑝𝑙𝑒 ′ 𝑠 𝑚𝑎𝑝𝑝𝑒𝑑 𝑟𝑒𝑎𝑑𝑠 𝑖𝑛 𝑚𝑖𝑙𝑙𝑖𝑜𝑛𝑠
Gene expression levels
Unannotated genes
Demultiplexing
Mapping
Removing duplicates and
non-unique mappings
Gene expression levels
Differential gene
expression
Differential expression parameters:
• Threshold - Minimum number of reads for pair
testing
• Normalization
• Replicates
Differential expression parameters affect
the results
RNA-Seq results
Coverage
Coverage – resequencing
Number of bases that cover each base in the
genome in average.
𝒂𝒗𝒆𝒓𝒂𝒈𝒆 𝒄𝒐𝒗𝒆𝒓𝒂𝒈𝒆 =
𝑟𝑒𝑎𝑑 𝑙𝑒𝑛𝑔𝑡ℎ ⋅ 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑟𝑒𝑎𝑑𝑠 ⋅ % 𝑢𝑛𝑖𝑞𝑢𝑒𝑙𝑦 𝑚𝑎𝑝𝑝𝑒𝑑 𝑟𝑒𝑎𝑑𝑠
𝑔𝑒𝑛𝑜𝑚𝑒 𝑠𝑖𝑧𝑒
“Coverage” – RNA-Seq
• Depends on the expression profile of each
sample.
• Highly expressed genes will be detected with less
“coverage” than lowly expressed genes.
What should you ask yourself before sequencing
when planning the experiment
• Reference genome:
–
–
–
–
What is my reference genome?
Does it have updated annotations?
What annotations are known?
Are my samples closely related to the reference genome?
• Do I expect to have contaminations in my sample?
• Do I have validations from other technologies? (RT-PCR,
SNPchip…)
• Do I have controls and replicates?
• RNA-Seq: am I interested in alternative splicing?
• Resequencing: What kind of mutations do I expect to find?