lecture 2 The first generation DNA Sequencing - cs.ucf.eduxiaoman/fall/lecture 2 The first generation DNA...DNA sequencing Determination of nucleotide sequence the determination of

The first generation DNA Sequencing

Slides 3‐17 are modified fromfaperta.ugm.ac.id/newbie/download/pak_tar/.../Instrument20072.pptslides 18‐43 are from Chengxiang Zhai at UIUC.

The strand direction

http://en.wikipedia.org/wiki/DNA

DNA sequencing Determination of nucleotide sequence

the determination of the precise sequence of nucleotides in a sample of DNA

Two similar methods:1. Maxam and Gilbert method

2. Sanger method

They depend on the production of a mixture of oligonucleotides labeled either radioactively or fluorescein, with one common end and differing in

length by a single nucleotide at the other end This mixture of oligonucleotides is separated by high resolution electrophoresis on polyacrilamide gels and the position of the bands

determined

Maxam-Gilbert

• Walter Gilbert– Harvard physicist– Knew James Watson– Became intrigued with

the biological side– Became a biophysicist

• Allan Maxam

The Maxam-Gilbert Technique

• Principle - Chemical Degradation of Purines– Purines (A, G) damaged by

dimethylsulfate– Methylation of base– Heat releases base– Alkali cleaves G– Dilute acid cleave A>G

Maxam-Gilbert Technique

• PrincipleChemical Degradation of Pyrimidines– Pyrimidines (C, T) are

damaged by hydrazine– Piperidine cleaves the

backbone– 2 M NaCl inhibits the

reaction with T

Advantages/disadvantagesMaxam-Gilbert sequencing

Requires lots of purified DNA, and many intermediate purification steps Relatively short readings

Automation not available (sequencers) Remaining use for ‘footprinting’ (partial protection against DNA modification when proteins bind to specific regions, and that produce

‘holes’ in the sequence ladder)

In contrast, the Sanger sequencing methodology requires little if any DNA purification, no restriction digests, and no labeling of

the DNA sequencing template

Sanger Method

Fred Sanger, 1958– Was originally a protein

chemist– Made his first mark in

sequencing proteins– Made his second mark in

sequencing RNA 1980 dideoxy

sequencing

Sanger Method

in-vitro DNA synthesis using ‘terminators’, use of dideoxy-nucleotides that do not permit chain elongation after their

integration DNA synthesis using deoxy- and dideoxynucleotides that

results in termination of synthesis at specific nucleotidesRequires a primer, DNA polymerase, a template, a mixture of

nucleotides, and detection system Incorporation of di-deoxynucleotides into growing strand

terminates synthesis Synthesized strand sizes are determined for each di-deoxynucleotide by using gel or capillary electrophoresis

Enzymatic methods

deoxyribonucleotide

Dideoxynucleotide

no hydroxyl group at 3’ endprevents strand extension

CH2O

OPPP5’

3’

BASE

CCGTAC3’ 5’5’ 3’primer

dNTP

ddATP

GGCA

ddTTP

GGCAT

ddCTP

GGC G

ddGTP

GGGGCATG

A T C G

Sample Output

1 lane

Phredhttp://www.phrap.org/phrap.docs/phred.html

Sanger sequencing• Laser excitation of fluorescent labels as fragments of discreet lengths 

exit the capillary, coupled to four‐color detection of emission spectra, provides the readout that is represented in a Sanger sequencing ‘trace’. Software translates these traces into DNA sequence, while also generating error probabilities for each base‐call.

• Simultaneous electrophoresis in 96 or 384 independent capillaries provides a limited level of parallelization.

• After three decades of gradual improvement, the Sanger biochemistry can be applied to achieve read‐lengths of up to ~1,000 bp, and per‐base ‘raw’ accuracies as high as 99.999%. In the context of highthroughput shotgun genomic sequencing, Sanger sequencing costs on the order of $0.50 per kilobase.

Comparison

• Sanger Method– Enzymatic– Requires DNA synthesis– Termination of chain

elongation

• Maxam Gilbert Method– Chemical– Requires DNA– Requires long stretches of

DNA– Breaks DNA at different

nucleotides

How to obtain the human genome sequence

• The Sanger sequencing can only generate 1kb long DNA segments.

• How to obtain the human genome that are 3 billion letters? The answer is to get pieces of DNA segments and assemble them into the genome.

Challenges with Fragment Assembly

• Sequencing errors

~1‐2% of bases are wrong

•Repeats

false overlap due to repeat

Bacterial genomes:5%Mammals: 50%

Repeat Types• Low-Complexity DNA (e.g. ATATATATACATA…)

• Microsatellite repeats (a1…ak)N where k ~ 3‐6(e.g. CAGCAGTAGCAGCACCAG)

• Transposons/retrotransposons– SINE Short Interspersed Nuclear Elements

(e.g., Alu: ~300 bp long, 106 copies)

– LINE Long Interspersed Nuclear Elements~500 ‐ 5,000 bp long, 200,000 copies

– LTR retroposons Long Terminal Repeats (~700 bp) at each end

• Gene Families genes duplicate & then diverge

• Segmental duplications ~very long, very similar copies

Strategies for whole‐genome sequencing

1. Hierarchical – Clone‐by‐clone yeast, worm, humani. Break genome into many long fragmentsii. Map each long fragment onto the genomeiii. Sequence each fragment with shotgun

2. Online version of (1) – Walking rice genomei. Break genome into many long fragmentsii. Start sequencing each fragment with shotguniii. Construct map as you go

3. Whole Genome Shotgun fly, human, mouse, rat, fugu

One large shotgun pass on the whole genome

Hierarchical Sequencing vs. Whole Genome Shotgun

• Hierarchical Sequencing– Advantages:  Easy assembly– Disadvantages:  

• Build library & physical map;          • Redundant sequencing

• Whole Genome Shotgun (WGS)– Advantages:  No mapping, no redundant sequencing– Disadvantages: Difficult to assemble and resolve repeats

Whole Genome Shotgun appears to get more popular… 

Whole Genome Shotgun Sequencing

cut many times at random

genome

forward-reverse paired readsknown dist

~500 bp~500 bp

Fragment Assembly

Cover region with ~7-fold redundancyOverlap reads and extend to reconstruct the

original genomic region

reads

Read Coverage

Length of genomic segment: GNumber of reads:  NLength of each read: L

Definition: Coverage  C = NL/ G

C

Enough Coverage

How much coverage is enough?

According to the Lander‐Waterman model:

Assuming uniform distribution of reads, C=7 results in 1 gap per 1,000 nucleotides

Lander‐Waterman Model• Major Assumptions

– Reads are randomly distributed in the genome– The number of times a base is sequenced follows a Poisson 

distribution

• Implications– G= genome length, L=read length, N = # reads– Mean of Poisson: =LN/G (coverage)– % bases not sequenced: p(X=0) =0.0009 = 0.09%– Total gap length: p(X=0)*G– Total number of gaps: p(X=0)*N

( )!

xep X xx

Average times

This model was used to plan the Human Genome Project…

Overlap‐Layout‐Consensus Assemblers: ARACHNE, PHRAP, CAP, TIGR, CELERA

Overlap: find potentially overlapping reads

Layout: merge reads into contigs and                   contigs into supercontigs

Consensus: derive the DNA sequence and correct read errors ..ACGATTACAATAGGTT..

Overlap

• Find the best match between the suffix of one read and the prefix of another

• Due to sequencing errors, need to use dynamic programming to find the optimal overlap alignment

• Apply a filtration method to filter out pairs of fragments that do not share a significantly long common substring

Overlapping Reads

TAGATTACACAGATTAC

TAGATTACACAGATTAC|||||||||||||||||

• Sort all k‐mers in reads      (k ~ 24)

• Find pairs of reads sharing a k-mer

• Extend to full alignment – throw away if not >95% similar

T GA

TAGA| ||

TACA

TAGT||

Overlapping Reads and Repeats

• A k‐mer that appears N times, initiates N2

comparisons

• For an Alu that appears 106 times  1012comparisons – too much

• Solution:Discard all k‐mers that appear more than 

t Coverage, (t ~ 10)

Finding Overlapping Reads

Create local multiple alignments from the overlapping reads

TAGATTACACAGATTACTGATAGATTACACAGATTACTGATAG TTACACAGATTATTGATAGATTACACAGATTACTGATAGATTACACAGATTACTGATAGATTACACAGATTACTGATAG TTACACAGATTATTGATAGATTACACAGATTACTGA

Finding Overlapping Reads (cont’d)

• Correct errors using multiple alignment

TAGATTACACAGATTACTGATAGATTACACAGATTACTGATAG TTACACAGATTATTGATAGATTACACAGATTACTGATAGATTACACAGATTACTGA

C: 20C: 35T: 30C: 35C: 40

C: 20C: 35C: 0C: 35C: 40

• Score alignments

•Accept alignments with good scores

A: 15A: 25A: 40A: 25-

A: 15A: 25A: 40A: 25A: 0

Multiple alignments will be covered later in the course…

Layout

• Repeats are a major challenge• Do two aligned fragments really overlap, or are they from two copies of a repeat? 

Merge Reads into Contigs

Merge reads up to potential repeat boundaries

repeat region

Merge Reads into Contigs (cont’d)

• Ignore non‐maximal reads• Merge only maximal reads into contigs

repeat region

Merge Reads into Contigs (cont’d)

• Ignore “hanging” reads, when detecting repeat boundaries

sequencing errorrepeat boundary???

ba

Merge Reads into Contigs (cont’d)

?????

Unambiguous

• Insert non-maximal reads whenever unambiguous

Link Contigs into Supercontigs

Too dense: Overcollapsed?(Myers et al. 2000)

Inconsistent links: Overcollapsed?

Normal density

Link Contigs into Supercontigs (cont’d)

Find all links between unique contigs

Connect contigs incrementally, if 2 links

Link Contigs into Supercontigs (cont’d)

Fill gaps in supercontigs with paths of overcollapsed contigs

Consensus

• A consensus sequence is derived from a profile of the assembled fragments

• A sufficient number of reads is required to ensure a statistically significant consensus

• Reading errors are corrected

Derive Consensus Sequence

Derive multiple alignment from pairwise read alignments

TAGATTACACAGATTACTGA TTGATGGCGTAA CTATAGATTACACAGATTACTGACTTGATGGCGTAAACTATAG TTACACAGATTATTGACTTCATGGCGTAA CTATAGATTACACAGATTACTGACTTGATGGCGTAA CTATAGATTACACAGATTACTGACTTGATGGGGTAA CTA

TAGATTACACAGATTACTGACTTGATGGCGTAA CTA

Derive each consensus base by weighted voting