DNA Sequencing and the Human Genome Project

1
(No Transcript)
2
DNA Sequencing and the Human Genome Project

• History
• Technology
• Analysis

3
Technology
4
Aspects of Sequencing Genomes

• Sequencing method
• Cloned DNA
• Clone/Sequence Assembly

5
Sequencing Methods

• Sanger chain termination method gt90 of all
  sequencing
• Relies on ability of DNA polymerase to
  incorporate nucleotide analogs while synthesizing
  template driven DNA

6
Dideoxynucleotide-based Sequencing
7
Automating Sanger Sequencing
8
Other Automated Methods

• Hybridization method
• Hybridize to oligos on a chip
• Affymetrix can do 30K resequence
• Limited by number of features and hybridization
  specificity
• Single molecule methods
• Pore-base - threads DNA through molecular pore in
  membrane - bases determined by changes in
  conductance
• Mass spec - best for small molecules now like SNPs

9
Most-used hardware

• ABI 377 - gel based - 96 lanes a pop - read
  length 500bp - run time 4-16h gt 40,000
  bases/run X 3 runs/day 120,000
• ABI 3700 - Capillary based - 48 capillaries -
  read length 500bp - run time 40 minutes gt
  950,000

10
Whaddaya determine the sequence of?

• Major problem is all methods only get 500bp/read
• Shotgun method can help

11
Basic Shotgun Strategy
12
Power of Shotgun

• Needs no prior knowledge of target
• Requires no maps or landmarks
• Got genome? Can start!

13
Limits of Shotgun

• Requires redundant sequence
• Not only bad - gives higher accuracy
• Requires representative library
• No missing clones
• Difficulty of locating overlaps
• Harder with larger genomes
• Vertebrate genomes have repetitive DNA
• Especially human 50 repeat

14
Computers Help

• When first begun
• PDP-11 (DEC) was state of art
• No PCs existed
• Data input punch cards
• Data output unformatted text!
• Conclude You guys are fortunate!

15
Locating Overlap - considerations

• Need rules
• How much overlap to call it real?
• How much mismatach in overlap?
• How to determine error?
• How to automate?
• Do you let the CPU do it?
• Do you Edit? If so how?

16
The CONTIG

• CONTIG CONTIGuous sequence from overlapping
  consensus
• Phrase coined by Roger Staden MRC
• First level of assembly in shotgun
• Definition of consensus carefully controlled

17
Sample Contig
GGCTCTTAGGAGATT

GATTTAGTTATGTTATTGTGCAACTATC

Overlap?
ATGTTATTCTGCAACCATCGCTGCGGACGAATAGCTGT

TTGTGCAACAATCGCTGCGGACGA
11111111111345224455662223333311111112233332333111
3333333
GGCTCTTAGGAGATTTAGTTATGTTATTGTGCAACNATCGCTGCGGACGA
ATAGCTGT
What constitutes consensus?
18
PHRED, PHRAP, CONSED, FINISHER

• Phils Rapid Editor (PHRED)
• Reads tracings from ABI
• Calls bases using best available graphical
  analyzer
• Makes quality assessment based on signal
  strength, background, overlapping bands etc.
• This gives a quantitative basis for
  establishing a contig

19
Phrap

• Phils Rapid Assembly Program
• Takes PHRED output as input
• Compares all PHRED reads to all other reads and
  contigs and -
• Makes tentative contigs with biases
• End overlaps better than middle
• Overlap must reach threshold score
• Score is identity plus PHRED quality factor

20
Consed

• CONtig Sequence Editor
• Permits finisher to edit overlaps
• Permits/confirms contig joins
• Permits (but discourages) sequence editing
• Allows identification of repeats
• Uses RepeatMasker output

21
Finisher

• Part of Consed
• Makes suggestions for closure
• Tells which clones to extend or reverse sequence
• Derives PCR primers for gap filling
• Estimated that finishing takes over twice as long
  as sequencing

22
Workflow
BAC or Small Genomic DNA
Fragment - sonication preferred
Clone library - 5-10X representative
Sequence - Enough runs
Data to PHRED -gt PHRAP -gt Consed -gt Finisher
Decide to fill gaps or done
Post to NCBI
23
Strategies

• Divide and conquer
• Create physical map
• Create smaller and smaller subclones of mapped
  pieces
• Carry out shotgun sequencing smallest pieces
• Whole genome
• Generate sequence-able clones
• Determine sequences at random using shotgun
• Use sequence overlap to reassemble into consensus

24
Philosophical issues regarding which to use
Wet-bench intensive
Fully map then sequence
Partial sequence ends, construct map, then
sequence
Full genome shotgun
Computationally intensive
25
Limitations on shotgun sequencing genomes

• Obtaining enough clones to cover all spots
• Finding credible sequence overlaps
• Repetitive DNA in Humans
• Computational power

26
More on repeats
Reads are only 500bp
GCTAGGCTAGTGGCATG
Genome is 3,000,000,000bp
Identical repeat sequences are interspersed throug
hout the genome so impossible to
place repeat-containing reads.
Reads are only 500bp
GCTAGGCTAGTGGCATG
GCTAGGCTAGTGGCATG
GCTAGGCTAGTGGCATG
GCTAGGCTAGTGGCATG
GCTAGGCTAGTGGCATG
Clone 1 sequence
CGAGCGTGTTGTACGTGTGA
GCTAGGCTAGTGGCATG
Clone 2 sequence
GGAGTGCTGAGTGGTGCAGCTAGGCTAGTGGCATGGGAGTGCTGAGTGGT
GCA
27
Mapping first - Shotgun sequence later
M13 or plasmid - 1 BAC (150kb) needs 6000
sequence reads or 2-3000 clones
28
Clones

• Large insert clones
• YACs (Yeast Artificial Chromosomes
• Useful for mapping 1mb inserts
• Unstable during construction and propagation
• Not useful for sequencing
• BACs (Bacterial Artificial Chromosomes)
• 150kb insert
• Extremely stable and easy to propagate
• Gold standard for sequencing targets and
  chromosome-scale maps
• Cosmids
• 50kb insert
• Extremely stable and easy to propagate
• Useful for sequencing but too small for
  chromosome maps

29
Sequence-ready clones

• Plasmids
• 1-10kb insert capacity
• High copy number
• Easy to sequence bi-directionally
• Automated clone picking/DNA isolation possible
• Examples pUC18, pBR322
• Single-stranded Bacteriophage
• 1-5kb insert capacity
• Grows at high copy as plasmid and is shed into
  medium as single stranded DNA phage
• Easy to isolate, pick, sequence
• Easy to automate
• M13 is used almost exclusively

30
Mapping

• Human Genome Maps
• BAC Fingerprint map
• Genetic Map
• Cytogenetic Map
• STS-based physical map (YACs)
• Radiation Hybrid Map

31
Clone map from USSC
32
Genetic Map

• Genethon and Marshfield
• Used CEPH families to map
• Used microsatellite markers (highly polymorphic)
• Mapping on only 100 families attained 0.7cM map
• Gave 5000 well ordered PHYSICAL markers
• Can be used to order clones and contigs

33
Cytogentics

• FISH (Fluorescence in situ hybridization) -
  useful to locate clones

34
STS Content YAC Map
35
RH Map
36
RH principle
37
General RH mapping on Panels
38
RH Mapping Panels

• Genebridge
• Number 93
• Retention 32
• Avg. Size 25mb

• Stanford G3
• Number 83
• Retention 16
• Avg. Size 2.4mb

39
(No Transcript)
40
Output from Stanford

• From rhserver_at_paxil.stanford.edu
• Date Tue Sep 9, 2003 92726 AM America/Denver
• To krauter_at_colorado.edu
• Subject SHGC RHSERVER
• This email message has been sent automatically by
  the StanfordHuman Genome Center RHserver in
  response to your
• submission.If you have questions or comments
  please submit them towebmaster_at_shgc.stanford.edu
  and include the
• message ID krauter_at_colorado.edu1063121245.93
• Duplicate markers are indicated with a (D) after
  the marker name,a LOD score is now given for
  duplicates.Reference
• Number Stanford RH Panel G3 Lowest LOD
  Reported 4 Chromosome Value 0
• Results for HUM_GEN
• ----------------------------------------
• SubmittedVector1100000010000000101000100000100000
  0110011001000001100100010000011101000
• 110000000110
• SHGCNAME CHROM
  LOD_SCOREDIST. (cRs)
• 1 SHGC-57080 22
  19.18 4
• Vector1100000000000000101000100000100000011001100
  10000011001000100000111010001100000001102
• 2 SHGC-7822 22
  18.63 4
• Vector1100000000000000R01000100000100000011001100
  10000011001000100000111010001100000001103
• 3 SHGC-58507 22
  17.15 7

41
END of LINE

• At the end, one has
• Detailed marker and clone maps
• Collection of BACs covering most of genome
• Sequence minimal tiling path of BACs

42
Assembly-line process at MIT Genome Center
Grow in 2ml cultures
Pick from plate into dishes
Bar code 384-well dishes
ABI 3700 Sequencer
Multiposition robot preps DNA
Sanger rxns done in thermal cyclers
43
Technical sidebar

• Literally hundreds of millions of clones must be
  sequenced
• Must automate (i.e. use robots)
• Methods to pick clones with inserts, prepare DNA,
  carry out sequencing reactions and load automated
  sequencers must be fully automatic (no human
  steps)

44
Both Celera and HGC compromised

• Human Genome Consortium
• Derived STS maps
• Sequenced BAC ends and fingerprinted to make maps
• Then sequenced minimum tiling path of BACs
• Celera
• Did full random shotgun of 1-3kb and 10-20kb
  clones
• Used STS, EST, and BAC maps to order small
  contigs into larger contigs

45
Whats the diff?

• Did the methods produce different outcomes?
• No
• Both produced gapped sequences
• Both lacked highly repeated segments
• Both produced sequence of sufficient quality to
  begin detailed analyses
• Yes
• Several regions had significantly different
  sequence orders
• Not all genes in one were present in the other
• HGC had, on average smaller but better contigs
• Celera had higher redundancy (i.e. accuracy)
  sequence

46
Products

• HGC
• Reads 181 X 106
• Bases 23 X 109
• Av. Contig 3 X 105
• No. Gaps 1.5 X 105
• No. Genes 24,500

• Celera
• Reads 27 X 106
• Bases 14 X 109
• Av. Contig 3 X 106
• No. Gaps 1 x 105
• No. Genes 26,383