Sean M. Sykes

2007 Finishing In The FutureSanta Fe, NM06/19/2007

Incorporating New Sequencing Technologies Into Finishing Strategy

Overview

• Problems With Conventional Methods

• Roche/454 Brings Something New To The Table: Both Good And Bad

• What Can We Do About 454 Limitations?

Neurospora Finishing

39.2 Mb Finished Quality Sequence- > 20x Sequence Coverage- > 100x Physical Coverage

- 8 library types

41 Mb In Optical MapNeurospora Chromosome 2

Neurospora Gaps In “Finished” Assembly

~180 Gaps Remain- Unclonable- Low % GC

GC75%

50%

25%

280Kb 380300 320 340 360

GC Distribution (120 kb)

What Does 454 Provide?

24x 454 Assembly41.5 Mb Total Length5017 Contigs

Adds ~2 Mb of new sequence

Does 454 Improve Neurospora Assembly?

GC75%

50%

25%

280Kb 380300 320 340 360

41.5 Mb 24x 454 AssemblyFinished Territory: 44% GC

• Spans 36 gaps• Extends 200 additional ends

454 Only Territory: 27% GCGC75%

50%

25%

280Kb 380300 320 340 360

Who Else Needs Help?

ABICoverage 8X% Genome 80%Contigs 315

Listeria monocytogenes2.9 Mb Genome

• Typical 8x draft assembly covers >= 95%• Significant cloning bias here

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

454 Gets Stuff ABI Misses

ABI 454Coverage 8X 23X% Genome 80% 95%Contigs 315 51

Yes indeed!

Listeria story like Neurospora?

Yes- unclonable regions sequenced by 454No- not correlated to GC content

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

0

5

10

15

20

25

30

35

40

45

50

g

0

2

4

6

8

10

12

14

16

18

SC

454 Captures Regions of Cloning Bias45

4 co

vera

ge

AB

I cov

erag

e

Coverage distribution (500 kb)

0

5

10

15

20

25

30

35

40

45

50

0 0 0 0 0 0 0 0 0 0 0 0 0 0

g

0

2

4

6

8

10

12

14

16

18

SC

Bias Does Not Correlate To GC

% G

C

-50

-40

-30

Coverage distribution (500 kb)

454 Works…End Of Story?

Not Really…

Some Questions Remain:

• Different Bias?

•Accuracy?

454 Coverage: A Closer Look

Human BAC 454 Coverage Across 100 bp GC Windows

0

10

20

30

40

50

60

70

80

90

100

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91

% GC

0

100

200

300

400

500

600

Win

do

ws

An

aly

zed

Windows

Human BAC 454 Coverage Across 100 bp GC Windows

454 Coverage: A Closer Look

0

10

20

30

40

50

60

70

80

90

100

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91

% GC

0

100

200

300

400

500

600

Win

do

ws

An

aly

zed

Avg. Coverage Windows

Human BAC 454 Coverage Across 100 bp GC Windows

Is Coverage Drop An Isolated Case?

0

25

50

75

100

125

150

175

200

225

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97 101

% GC

0

500

1000

1500

2000

2500

3000

3500

Win

do

ws

An

aly

zed

454.12 454.13 454.14 454.15 454.16 454.17 All Windows

• Different Bias?• Yes

Accuracy?

Revisiting 454 Questions

Assessing Sequence Accuracy of 454

Compare 454 assembly to finished sequence

• Base errors rare in consensus (~q40)• >95% of errors are insertion/deletion

• Majority in homopolymer runs • ~1% error rate in 5+ bp homopolymer runs

Hypothetical Broaderia Genome

• 3 Mb genome• 3000 genes • 1/10 kb Indel errors• 300 errors/genome

– 270 in coding region– frameshift in gene

9% gene models broken

Frameshifting Indels In Listeria

• Collaborators less than delighted• Validation showed only 8/41 correct

Total Percent 10403s 47 1.57J0161 130 4.33J2818 238 7.93

Gene Models Broken

Error Rates In Listeria 454 Assemblies

ABI q40

454 Assembly

Indel errors per 10kb in Listeria

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

10403S (24x) J0161 (29x) J2818 (24x)Genome (Cover

9% GeneModels Broken

Increased 454 Coverage Increase Accuracy?

Errors per 10kb in 10403S Assem

0

200

400

600

800

1000

1200

1400

48x 36x 24x 12xCoverage

0%

10%

20%

30%

40%

50%

60%

70%

80%

% 4

54 A

ssem

bly

Cov

ered

by

AB

I

Indel Error Rate Genome Covere

• Different Bias?• Yes

• Accuracy?• Indels break gene models

Revisiting 454 Questions

Solexa/Illumina Improve 454 Accuracy?

Solexa Accuracy

High sensitivity and specificity

Sequenced 1 lane of F11 by Solexa2.5M passing reads, 27b = ~14x coverage

Aligned all reads to H37Rv genome

Identified 98% of 800 known SNPsIdentified 95% of indelsNO false positivesMissing SNPs in regions >80%GC

Sequence Tb by SolexaCompare to finished sequence

Indel Errors per 10kb in Listeria Genom

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

10403S (24x) J0161 (29x) J2818 (24x)Genome (Coverage

Pre Solexa Correction After Solexa Correctio

ABI q40

Can Solexa Fix 454 Errors In Listeria?

454 Assembly

Solexa

Conclusions

• 454 captures regions of ABI cloning bias, but other biases still exist• 454 indels impact gene annotation• Solexa data correct indel errors in 454 data• Significant improvement in consensus quality for relatively small cost

Acknowledgements

Assembly Analysis Team• Sarah Young• Aaron Berlin

WGA Team• Will Brockman• Pablo Alvarez• David Jaffe

Chad NusbaumBruce Birren

Technology Development• Georgia Giannoukos• Will Lee• Sheila Fisher• Jim Meldrim• Nabil Hafez• Diana Shih• Tracey Honan• Todd Sparrow• Michael Weiand• Zac Zwirko• Nicole Stange-Thomann• Alex Melnikov• Andreas Gnirke• Pat Cahill• Robert Nicol

Computer Finishing• Mike FitzGerald• Lynne Aftuck

Genome Annotation• Chandri Yandava• Chinnapa Kodira

Scientific Project Mgt• Carsten Russ