Whitehead Institute for Biomedical Research

4/20/11 Yaniv ErlichDNA Sudoku

Yaniv Erlich

Exome sequencing and disease-network analysis

Whitehead Institute for Biomedical ResearchWhitehead Institute for Biomedical Research

Two projects with neurological diseases

yaniv@wi.mit.edu

1. Joubert syndrome in Ashkenazi Jews

2. Hereditary Spastic Paraparesis in Palestinians

What is Joubert Syndrome?

• Cerebello-oculo-renal phenotype:

- Hypoplasia of the cerabellar vermis

- Pyschomotor retardation and hyptonia

- Extra digits in upper and lower limbs (sometimes)

- Lazy eye

- Renal insufficiency

• Molar Tooth Sign in MRI:

yaniv@wi.mit.eduExome sequencing and disease network

Parisi, 2007

Normal Joubert

Cerebellum

Thick and elongated superior cerebellar

peduncles

Large Interpeduncular

fossa

Intro. Our approach Results Conclusion

The cases

• Dor-Yeshorim identifies 13 cases in Jewish Ashkenazi families

• 8 families, where 3 are part of the same clan.

• Autosomal recessive pattern

• Founder effect

yaniv@wi.mit.eduExome sequencing and disease network

Intro. Our approach Results Conclusion

The starting point

• 9 genes are known to cause Joubert syndrome in other populations

• Sequencing those genes revealed normal results

• Social implication…

• Using autozygoustity mapping Hadassah found a strong signal from the centromeric region of chromosome 11.

yaniv@wi.mit.eduExome sequencing and disease network

Chr 11

Hundreds of exons…

Intro. Our approach Results Conclusion

Finding the mutation – our approach

• Sequence the entire exome of a healthy mother and an affected daughter.

• Finding mutations that are:

- Heterozygous in the mother

- Homozygous in the child

- Not in dbSNP

- Causing a change in the coding region

• Two lanes 36x2 per specimen

• Hadassah sequenced the exons in the region according to a prioritized list of candidate exons.

yaniv@wi.mit.eduExome sequencing and disease network

Intro. Our approach Results Conclusion

Technical details

yaniv@wi.mit.eduExome sequencing and disease network

Step Tool Comments

Sequencing Illumina Mother: ~67,400,000 single end-reads

Child: ~73,600,000 single end-reads

Alignment Bowtie - Default parameters -> poor results. (only 1% of reads were mapped).

- Iterative mapping

Total unique mapper:

Mother – 51 million (76%)

Child – 54 million (74%)

SNP calling SOAPsnp Only regions with x5 coverage or more

Intro. Our approach Results Conclusion

Analysis of variations

yaniv@wi.mit.eduExome sequencing and disease network

SNPs in: #

Mother 49,515

Child 48,142

That are shard… 23,986

And are heterozygous in mother and

homozygous in child… 2,541

And not in dbSNP 105

And are not synonymous 39

And are in the mapped locus 1

Intro. Our approach Results Conclusion

And the winner is… TMEM216

yaniv@wi.mit.eduExome sequencing and disease network

Amino-acid change: Arg -> Leu

(CGC>CTC)

Chromosome 11

Intro. Our approach Results Conclusion

Who are you TMEM216?

yaniv@wi.mit.eduExome sequencing and disease network

• Transmembrane protein

• 88 amino-acid

• Not a single paper on that gene

• Conservation analysis:

Intro. Our approach Results Conclusion

Additional lines of evidence

yaniv@wi.mit.eduExome sequencing and disease network

• PolyPhen predicated as damaging

•Hadassah found the same mutation (double blind)

- Found in all 13 cases and parents found as carriers.

Intro. Our approach Results Conclusion

From: Elpeleg OrlySent: Sun 12/6/2009 11:56 PMTo: Erlich, YanivSubject: RE: Preliminary analysis

BINGO________________________________________From: Erlich, Yaniv [erlich@cshl.edu] Sent: Sun 12/6/2009 23:32To: Elpeleg OrlySubject: Preliminary analysis

Hi Orly,

The only potential homozygous SNP mutation we found on chr11 between 59.5M-62M that is not in dbSNP and has an affect on the protein (missense, nonsense, splice) isTMEM216 Arg12->Leu (chr11:60918013).

Are we right?Thanks,Yaniv

Summary

yaniv@wi.mit.eduExome sequencing and disease network

• We found that TMEM216 is the causative mutation of Joubert syndrome in Ashkenazi Jews.

• The project took 2 months end-to-end in our side.

• We used only 4 lanes of Illumina GAII with paired-end 36nt reads.

• Israel Ministry of Health added the mutation to their test panel.

• Carrier rate in Ashkenazi Jews is 1:92

Intro. Our approach Results Conclusion

Two projects with neurological diseases

yaniv@wi.mit.edu

Intro. Our approach Results Conclusion

1. Joubert syndrome in Ashkenazi Jews

2. Hereditary Spastic Paraparesis in Palestinians

Hereditary Spastic Paraperesis

4/20/11 Yaniv ErlichExome sequencing and disease network

-A single Palestinian family:

-

- 3 brothers suffers from progressive weakness of the legs and abnormal gait

- Phenotype is HSP – degradation of the pyramidal tract.- 20 genes have been documented

(Weber J,2003)

Intro. The problem Our approach Results Conclusion

Thoughts on the Joubert rejection process

yaniv@wi.mit.eduExome sequencing and disease network

SNPs in: #

Mother 49,515

Child 48,142

That are shard… 23,986

And are heterozygous in mother and

homozygous in child… 2,541

And not in dbSNP 105

And are not synonymous 39

And are in the mapped locus 1

Currently, the rejection process is based on two classes of arguments:1.Genetic arguments2.Loss of function arguments

Genetic arguments

Loss-of-function arguments

Genetic arguments have weaker power with smaller familiesWe need a new class of arguments

Intro. The problem Our approach Results Conclusion

Disease-network analysis

yaniv@wi.mit.eduExome sequencing and disease network

Idea: Genes with a similar phenotype have a similar biological signature

Method: Stratify variations according to their similarity to known disease genes.

Zippi Brownstein, 2009

Intro. The problem Our approach Results Conclusion

Data gathering stage

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

Whole exome seq. with Illumina

Whole genome genotyping with Affy 250K array

Results

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

• 20 known genes that cause HSP were found intact

• None of the 40,000 disease causing mutations in HGMD were found in the exome sequencing data.

• X-linked mutation was ruled out.

• Genotyping identified 4 regions of homozygosity:

Conclusion: it is a new gene

Exclusion process

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

TotalGenes Position

SequencedGenes Position

Now what?

Disease network analysis rank KIF1A as the top candidate

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

Validation by loss of function analysis

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

TotalGenes Position

SequencedGenes Position

Supporting evidence

4/20/11 Yaniv ErlichExome sequencing and disease network

Intro. The problem Our approach Results Conclusion

• Sanger sequencing confirmed the mutation in third affected child. 4 healthy brothers were not homozygous for the mutation

• KIF1A is a kinesin. Phenotype is neuronal.

• The region of KIF1A was suspected to cause HSP by a previous study with multiple Algerian family.

• Conservation suggests a functional domain in KIF1A

Summary

yaniv@wi.mit.eduExome sequencing and disease network

• Identifying a new gene for spastic paraperesis.

• Using only a single family and a new class of arguments

•The rate of the mutation is 1:200 in Palestinians.

Intro. The problem Our approach Results Conclusion

yaniv@wi.mit.edu

Erlich LabWhitehead Institute

Positions available (yaniv@wi.mit.edu)