Next Generation Sequencing Data Analysis

1
Next Generation Sequencing Data Analysis

• Nadia Pisanti, University of Pisa

2
Why sequencing?

• The knowledge of DNA and RNA sequences has become
  a crucial tool for
• Basic research in biology, pharmacology and
  medicine.
• Many applied fields diagnostic (genetic diseases
  detection), pharmacogenomics (influence of
  genetic variation on drug response) and
  personalized medicine, forensic biology, gene
  therapies, biological systematics (the study of
  the diversification of living forms)

3
Sequencing some history

• "rapid DNA sequencing" by Frederick Sanger (UK)
  in the 1970s, became the method of choice for DNA
  sequencing, and was worth him his 2nd Nobel Prize
  in chemistry in 1980.
• Sanger method for sequencing DNA was used in the
  Human Genome Project (HGP) that produced the
  first reference sequence of the human genome.
• The HGP started in 1990 and was expected to take
  15 years.
• A first "rough draft" was finished in 2000 and
  announced in a press conference by Bill Clinton
  and Tony Blair!
• The complete genome was announced in 2003.
• Why announcing the rough draft in 2000?
• Why did the HGP take less than expected?

4
Celera Genomics- cut and paste from wikipedia
and my memory -

• In 1998, the American NIH researcher Craig Venter
  announced that his private company Celera
  Genomics would sequence the human genome at a
  fraction of the cost of the public project.
• A significant portion of the human genome had
  already been sequenced when Celera entered the
  field and was freely available to the public from
  GenBank.
• Celera used a technique called whole genome
  shotgun sequencing. This novelty spurred the HGP
  to change its own strategy, leading to a rapid
  acceleration of the public effort.
• Celera filed preliminary ("place-holder") patent
  applications on 6,500 whole or partial genes.
  Celera also promised to publish their findings in
  accordance with the terms of the 1996 "Bermuda
  Statement," by releasing new data annually (the
  HGP released its new data daily), although,
  unlike the publicly funded project, they would
  not permit free redistribution or scientific use
  of the data. For this reason, the public
  competitor was compelled to publish the first
  draft of the human genome before Celera.
• In 2000, the HGP released a first working draft
  on the web. The scientific community downloaded
  one-half trillion bytes of information from the
  UCSC genome server in the first 24 hours of free
  and unrestricted access to the first ever
  assembled blueprint of our human species.
• Also in 2000, president Clinton announced that
  the genome sequence could not be patented, and
  should be made freely available to all
  researchers. The statement sent Celera's stock
  plummeting and dragged down the
  biotechnology-heavy Nasdaq. The biotechnology
  sector lost about 50 billion in market
  capitalization in two days. But the public
  release of the data ensured its fair use and
  availability.

5
shotgun sequencing

• The Sanger sequencing technology could only be
  used for short DNA fragments (from 100 to 1000
  bases) DNA must thus be divided into small
  pieces, and then be re-assembled.
• This can be done in two ways
• Chromosome walking sequencing piece by piece
  consecutive fragments.
• Shotgun sequencing break several copies of the
  DNA strand into random overlapping fragments,
  sequencing them, and then re-assemblying in
  silico exploiting the overlap.

• Since when shotgun sequencing was introduced by
  Celera, it is the method of choice for large
  scale sequencing.

6
Shotgun sequencing assembly

• Wikipedia, about shotgun sequencing "faster but
  more complex".
• The "complexity" of the approach is because of
  algorithmic issues
• (Eu)gene Myers, a string algorithms expert, was
  leading the computer scientists at Celera he
  made the difference
• Challenges in assembly phase finding
  prefix/suffix overlap, data structure for storing
  fragments and "overlap graph", assembly algorithm
  managing duplications.

7
Fragment Assembly
The problem of sequence assembly can be compared
to taking many copies of a book, passing them all
through a shredder, and piecing the text of the
book back together just by looking at the
shredded pieces. Besides the obvious difficulty
of this task, there are some extra practical
issues the original may have many repeated
paragraphs, and some shreds may be modified
during shredding to have typos. Excerpts from
another book may also be added in, and some
shreds may be completely unrecognizable.
8
What is NGS?

• Next/New Generation Sequencing
• Massively Parallel Sequencing
• Third Generation Sequencing
• High Throughput Sequencing

millions of fragments (reads) in a single run
!! by means of new technologies developed mainly
by

• Lynx Therapeutics merged with Solexa and they
  were bought by Illumina.
• ABI SOLiD
• ION Torrent Systems
• 454 Life Science acquired by Roche Diagnostics

they actually differ quite a lot on performances
and characteristics.
9
What's new with NGS?

• Sequencing the whole human genome took the HGP
• 3.000.000.000 dollars
• 13 years

• Sequencing a whole human genome now with NGS
  techniques takes
• about 1.000 dollars
• 4-5 days

Sequencing is much faster and (thus) cheaper !!
10
What is NGS great for

• re-sequencing no assembly, just mapping on a
  known reference genome.
• Metagenomics
• Transcriptome Sequencing RNA-Seq
• Chromatin immunoprecipitation combined with DNA
  sequencing ChIP-Seq

11
re-sequencing

• Sequencing a new individual of a species for
  which the reference genome is know (and (well)
  annotated).
• Important applications
• Medicine
• Building datasets of several strains of the same
  organism to investigate intra-species evolution.

12
re-sequencing medical applicationswe will get
back to this later

• Genotyping testing for known mutations
  (sequencing can be possibly targeted to specific
  regions).
• Variation analysis scanning for any mutation
  such as Single Nucleotide Polymorphisms (SNPs),
  or Copy Number Variations (CNVs) or other
  Structural Variants (SVs) that can be associated
  to congenital diseases, predisposition for
  certain pathologies, or drug response.
• Most of NGS tools offer the relative software to
  detect mutations.
• With NGS these tests can be made on large scale.
  and back in time Roche sequenced the Neanderthal
  genome in 2006!

13
re-sequencing

• Challenges for computer science
• Indexing data and (quickly) mapping on reference
  genome
• SNPs and SVs calling.
• Mind the repeats up there!

• Challenges for informatics
• Build tools for genetists.
• Interpreting SNPs and SVs crossing with DB
  information.
• DB management

14
metagenomics

• Metagenomics essentially entails brute force
  sequencing of DNA fragments obtained from an
  uncultured, unpurified, microbial and/or viral
  population, followed by bioinformatics-based
  analyses that attempt to answer the question
  "Who's there?" E.R.Mardis, Trends in genetics
  2008

• Characterizing the human microbiome we live in
  symbiosis with millions of microbial species.
  There is a theory saying that these symbiotic
  microbes provide an extension of the human genome
  and hence contribute to its genetic potentials in
  terms of protective immunity, added enzymatic
  capability
• Metagenomics not only in human body, but also in
  important ecosystems such as ocean, soil, deep
  mines.
• Metagenomics costs are effordable only now with
  NGS (mostly 454 Roche as with longer reads they
  better allow de novo sequencing)

15
What is RNA-Seq

• NGS opened a new phase
• in transcriptomics (aka
• expression profiling)
• thanks to
• low requirements of
• nucleotide sequence
• product
• and
• deep coverage

16
Why RNA-Seq

• Among the goals of the HGP there was the mapping
  and genotype associated to (the predisposition
  for) diseases.
• It is now very clear (and it was not then) that
  reading the genome is not enough
• Same genome, different phenotypes and different
  diseases how comes?
• Environmental effects (food, pollution, life
  style) act on gene transcription.
• We ought to investigate the transcriptome!
• The transcriptome are the genes that are being
  actively expressed at a given time.
• The role of miRNA for gene regulation.

17
RNA-Seq

• Sequencing the transcriptome to investigate
  differentially expressed genes
• under different conditions, or
• in different tissues
• in different alleles
• The different expression can be in quantitative
  terms or in alternative splicing terms
  (eukaryotes only).

de novo transcriptome assembly
18
RNA-Seq

• Sequencing the transcriptome to investigate
  differentially expressed genes
• under different conditions, or
• in different tissues
• in different alleles
• The different expression can be in quantitative
  terms or in alternative splicing terms
  (eukaryotes only).

transcriptome re-sequencing
19
RNA-Seq quantification

• RNA-Seq (Quantification) is used to analyze gene
  expression of certain biological objects under
  specific conditions.

20
Alternative Splicingwe will get back to this
later

• AS is when several mRNAs can be produced from a
  unique pre-mRNA
• E.g. in humans there are approximately 30,000
  genes and it is estimated that 70 of human
  protein-coding genes undergo alternative splicing
  to generate up to 150,000-200,000 mRNAs and
  proteins through alternative splice site usage.
• In 2008, an experiment revealed that 34 of human
  transcripts were not from known genes Science
  321

21
non coding RNA

• ncRNA includes a wide class of regulatory RNA
  molecules whose function is as crucial as not yet
  understood.
• Discovering their sequences and (hence) genomic
  locations is hard because they (mostly) small and
  poorly conserved over evolutionary time.
• In silico prediction methods are of high
  importance and very promising, but so far of
  little use.
• Currently, ncRNA are mostly discovered by
  sequencing small RNA fragments, for which task
  NGS tools are ideal!
• In silico analysis of such data will be crucial
  for understanding it (secondary structure
  prediction, putative functions prediction based
  on learning methods).
• A new class of miRNA (or small RNA) is being
  discovered every day

22
ChIP-Seq

• ChIP-seq combines chromatin immunoprecipitation
  (ChIP) with massively parallel DNA sequencing to
  identify the binding sites of DNA-associated
  proteins.
• The goal is to analyze protein interactions with
  DNA (e.g. how transcription factors, that are
  proteins, regulate gene expression).

23
The bad side of NGS

• Even shorter fragments from 1000 of Sanger
  technology to 25, then 50, then 75, now 100
  bases.
• Even more errors (when new size is released).

Fragment assembly is even harder !!
24
From M.L. Metzker "Sequencing technologies the
next generation", Nature Reviews Genetics 11,
31-46, 2010
What is the best depends on what you need it
forand how much money you have
25
Roche 454 Genome Sequencer

• It was the first introduced in the market in
  2005.
• Its technology allows to produce relatively long
  reads (400-700 bases).
• Its base calling cannot handle long (gt6)
  stretches of the same nucleotide, resulting in
  insertions and deletions errors there
• On the other hand very low substitutions error
  rate.
• Overall error rate at 1.

26
Illumina Genome Analyzer(aka Solexa sequencer)

• The most widely available NGS technology.
• Reads up to 100b long.
• Error rate at 1-1,5, mostly substitutions
  (indels are much less common).

27
ABI's SOLiD

• Probably the second most widely used.
• The workflow is similar to Solexa/Illumina's.
• An interesting difference SOLiD uses a di-base
  sequencing technique in which two nucleotides are
  read simultaneously. 16 di-bases still
  represented by 4 "colors", but the one-base-shift
  solves the redundancy.
• As a consequence
• Sequencing error may propagate.
• Read alignment can be speed up.

• Error rate around 2-4

28
Paired-end and Mate-pairs

• Two very different objects from the point of view
  of the technology as they are obtained with very
  different procedures.
• Available from all NGS platforms.
• From the computational point of view, they are
  the same two sequences at an approximatively
  know distance from eachother in the genome
  (insert size).
• They are crucial to
• Correctly map/assemble repeated fragments
• Detect Structural Variants and Copy Number
  Variations.

29
Fragment Assemblywith NGS data

• It is like a diabolic sudoku
• - with very few initial numbers
• - many solutions satisfy the constraint choice
  is arbitrary
• - only one of the many solution is the good one,
  and there is no clue on which

30
NGS and Informaticsthe challenges 1

• Massive Image processing and basecalling within
  sequencing technology.
• Growing need of managing big data
• Indexing issues.
• Efficient mapping and alignments.
• Parallel and High Performance computing.
• New emphasis on efficient data structures and
  algorithms with special care on memory usage.

31
NGS and Informaticsthe challenges 2

• Designing and producing tools for data analysis
  integrating information from different sources
  (e.g. genome browsers).
• Designing and producing tools for assemblying..
• Designing and producing tools for genotyping a
  new one every day, hard to compare...
• Customized analysis informatics is needed for
  any project and in any lab.
• "Curiously" back to old style stuff such as
  command line, machine language programming