Phenotype analysis in humans using OMIM

Transcription

1 Outline: 1) Introduction to OMIM 2) Phenotype similarity map 3) Exercises Phenotype analysis in humans using OMIM Rosario M. Piro Molecular Biotechnology Center University of Torino, Italy 1 MBC, Torino May 26th 2007

2 What is OMIM? Hamosh et al., Nucleic Acids Res. 30(1):52-55, 2002 comprehensive knowledge-base of human genes and genetic disorders started by V.A. McKusick in 1966 as MIM available online since Online Mendelian Inheritance in Man maintained since 1995 at National Center for Biotechnology Information (NCBI) integrated with Entrez daily updated by Johns Hopkins University MBC, Torino May 26th 2007

3 OMIM entries (1) OMIM entries: currently 17,663 entries (May 14th, 2007) full-text summary of gene and/or phenotype links to other genetic databases, PubMed references, etc. MIM numbers/ids: six digits starting with: (100000(200000(300000(400000(500000( important fields: titles/symbols, gene map locus (cytogenetic location), text, clinical synopsis and features of a disorder, allelic variants, references,... (not all present) ) ) ) ) ) ) Autosomal dominant loci or phenotypes (entries created before May 15, 1994) Autosomal recessive loci or phenotypes (entries created before May 15, 1994) X-linked loci or phenotypes Y-linked loci or phenotypes Mitochondrial loci or phenotypes Autosomal loci or phenotypes (entries created after May 15, 1994) allelic variants: 10-digit number: <MIMentry>.NNNN MBC, Torino May 26th 2007

4 OMIM entries (2) Entry classes: the prefix of a MIM number indicates the type of entry: e.g , BREAST CANCER 1 GENE; BRCA1 Symbol * # + % <none> ^ 4 Category gene of known sequence descriptive entry, usually of a phenotype; does not represent a unique locus contains the description of a gene of known sequence and a phenotype confirmed Mendelian phenotype or phenotypic locus; molecular basis unknown phenotype with suspected Mendelian basis or independence of entry unclear entry no longer exists (removed or moved to another entry) MBC, Torino May 26th 2007

5 OMIM entries (3) Example: , BREAST CANCER 1 GENE; BRCA BREAST CANCER 1 GENE; BRCA1 Alternative titles; symbols BREAST CANCER, TYPE 1, INCLUDED BREAST CANCER 1, EARLY-ONSET, INCLUDED BREAST-OVARIAN CANCER, INCLUDED Gene map locus 17q21 TEXT For a general discussion of hereditary breast cancer, see CLINICAL FEATURES Familial Breast Cancer Features characteristic of familial, versus sporadic, breast cancer are younger age at diagnosis, frequent bilateral disease, and frequent occurrence of disease among men Hall et al. (1990). According to the conclusions of the Breast Cancer Linkage Consortium (1997), the histology of breast cancers in women predisposed by reason of carrying BRCA1 and BRCA2 (600185) mutations differs from that in sporadic cases, and there are differences between breast cancers in carriers of BRCA1 and BRCA2 mutations. [...] 5 MBC, Torino May 26th 2007

6 OMIM entries (3) Example: , BREAST CANCER 1 GENE; BRCA1 # BREAST CANCER BREAST CANCER 1 GENE; BRCA1 Alternative titles; symbols Alternative titles; symbols BREAST CANCER, TYPE 1, INCLUDED BREAST CANCER 1, EARLY-ONSET, INCLUDED BREAST-OVARIAN CANCER, INCLUDED Gene map locus 17q21 BREAST CANCER, FAMILIAL BREAST CANCER, FAMILIAL MALE, INCLUDED Gene map locus 22q12.1, 17q22-q23, 17q22, 17p13.1, 15q15.1, 13q12.3, 12p12.1, 11q22.3, 11p15.5, 8q11, 16p12, 3q26.3, 2q34-q35 TEXT TEXT For a general discussion of hereditary breast cancer, see A number sign (#) is used with this entry because of evidence that mutation at more than one locus can be involved in different families or even in the same case. These loci include BRCA1 (113705) on 17q, BRCA2 (600185) on 13q12, [...] CLINICAL FEATURES Familial Breast Cancer Features characteristic of familial, versus sporadic, breast cancer are younger age at diagnosis, frequent bilateral disease, and frequent occurrence of disease among men Hall et al. (1990). According to the conclusions of the Breast Cancer Linkage Consortium (1997), the histology of breast cancers in women predisposed by reason of carrying BRCA1 and BRCA2 (600185) mutations differs from that in sporadic cases, and there are differences between breast cancers in carriers of BRCA1 and BRCA2 mutations. [...] 6 MBC, Torino May 26th 2007

7 OMIM entries (3) Example: , BREAST CANCER 1 GENE; BRCA1 # BREAST CANCER BREAST CANCER 1 GENE; BRCA1 Alternative titles; symbols Alternative titles; symbols BREAST CANCER, TYPE 1, INCLUDED BREAST CANCER 1, EARLY-ONSET, INCLUDED BREAST-OVARIAN CANCER, INCLUDED Gene map locus 17q21 BREAST CANCER, FAMILIAL BREAST CANCER, FAMILIAL MALE, INCLUDED Gene map locus 22q12.1, 17q22-q23, 17q22, 17p13.1, 15q15.1, 13q12.3, 12p12.1, 11q22.3, 11p15.5, 8q11, 16p12, 3q26.3, 2q34-q35 TEXT TEXT For a general discussion of hereditary breast cancer, see A number sign (#) is used with this entry because of evidence that mutation at more than one locus can be involved in different families or even in the same case. These loci include BRCA1 (113705) on 17q, BRCA2 (600185) on 13q12, [...] CLINICAL FEATURES Familial Breast Cancer Features characteristic of familial, versus sporadic, breast cancer are younger age at diagnosis, frequent bilateral disease, and frequent occurrence of disease among men Hall et al. (1990). According to the conclusions of the Breast Cancer Linkage Consortium (1997), the histology of breast cancers in women predisposed by reason of carrying BRCA1 and BRCA2 (600185) mutations differs from that in sporadic cases, and there are differences between breast cancers in carriers of BRCA1 and BRCA2 mutations. [...] 7 MBC, Torino May 26th 2007

8 OMIM entries (3) Example: , BREAST CANCER 1 GENE; BRCA1 # BREAST CANCER BREAST CANCER 1 GENE; BRCA1 Alternative titles; symbols Alternative titles; symbols BREAST CANCER, TYPE 1, INCLUDED BREAST CANCER 1, EARLY-ONSET, INCLUDED BREAST-OVARIAN CANCER, INCLUDED Gene map locus 17q21 BREAST CANCER, FAMILIAL BREAST CANCER, FAMILIAL MALE, INCLUDED Gene map locus 22q12.1, 17q22-q23, 17q22, 17p13.1, 15q15.1, 13q12.3, 12p12.1, 11q22.3, 11p15.5, 8q11, 16p12, 3q26.3, 2q34-q35 TEXT TEXT For a general discussion of hereditary breast cancer, see A number sign (#) is used with this entry because of evidence that mutation at more than one locus can be involved in different families or even in the same case. These loci include BRCA1 (113705) on 17q, BRCA2 (600185) on 13q12, [...] CLINICAL FEATURES Familial Breast Cancer BREAST CANCER 2 GENE; BRCA2 Features characteristic of familial, versus sporadic, breast cancer are younger age at diagnosis, frequent bilateral Alternative titles;disease, symbols and frequent occurrence of disease among men Hall et al. (1990). GENE; FANCD1 According to the conclusions of the Breast Cancer Linkage Consortium (1997), the histology offancd1 breast cancers in women CANCER, TYPE 2, INCLUDED predisposed by reason of carrying BRCA1 and BRCA2 (600185) mutations differs from that in BREAST sporadic cases, and there are differences between breast cancers in carriers of BRCA1 and BRCA2 mutations. [...] BREAST CANCER 2, EARLY-ONSET, INCLUDED [...] 8 MBC, Torino May 26th 2007

9 Gene and Morbid Map Morbid Map (maps disorders to genes): Disorder Breast cancer (1) Breast cancer (1) Breast cancer (3) Breast cancer 2, early onset (3) Breast cancer, (3) Breast cancer, (3) Breast cancer, (3) Breast cancer, (3) Breast cancer, 11:22 translocation associated (1) Breast cancer, ductal (2) Breast cancer, ductal (2) Breast cancer, early-onset, (3) Breast cancer, invasive intraductal (3) Breast cancer, lobular (3) Breast cancer, male, with Reifenstein syndrome (3) Breast cancer, somatic, (3) Breast cancer, somatic, (3) Breast cancer, sporadic (3) Breast cancer, type 3 (2) (?) Breast cancer-1 (3) Breast-ovarian cancer (3) OMIM Location 17p13.3 6q p15.2-p q12.3 3q q22-q23 11p p q23 Chr.13 1p36 17q22 1p32 16q22.1 Xq11-q12 12p12.1 8q11 17q21 13q21 17q21 17q21 Gene Map (maps genes to disorders): Location 17q21 9 Symbol(s) BCPR ESR1, ESR TSG101 BRCA2, FANCD1 PIK3CA PPM1D, WIP1 SLC22A1L, BWSCR1A, IMPT1 TP53, P53, LFS1 BRCATA BRCD1 BRCD2 BRIP1, BACH1, FANCJ RAD54L, HR54, HRAD54 CDH1, UVO, LCAM, ECAD AR, DHTR, TFM, SBMA, KD, SMAX1 KRAS2, RASK2, NS3 RB1CC1, CC1, KIAA0203 PHB BRCA3, BRCAX BRCA1, PSCP BRCA1, PSCP Symbol Title MIM # BRCA1, PSCP Breast cancer-1 gene Disorder Comments Breast cancer-1 (3); Ovarian cancer (3); Breast-ovarian cancer (3); Papillary serous carcinoma of the peritoneum (3) MBC, Torino May 26th 2007 Method Fd, REc Mouse 11(Brca1)

10 Database Searches 10 Database searches: by MIM number by disorder or gene name/symbol plain English (e.g. tissues) MBC, Torino May 26th 2007

11 Example: OMIM search (1) 11 MBC, Torino May 26th 2007

12 Example: OMIM search (2) Filters (no filtering if not ticked) 12 MBC, Torino May 26th 2007

13 Example: OMIM search (2) 13 MBC, Torino May 26th 2007

14 Example: OMIM search (2) 14 MBC, Torino May 26th 2007

15 OMIM Output Formats (1) ASN.1 can also be saved as a file 15 MBC, Torino May 26th 2007

16 OMIM Output Formats (2) XML can also be saved as a file 16 MBC, Torino May 26th 2007

17 Example 2: Morbid Map 17 MBC, Torino May 26th 2007

18 Example 3: Gene Map 18 MBC, Torino May 26th 2007

19 OMIM Downloads 19 MBC, Torino May 26th 2007

20 Phenomics systematic classification/grouping of relationships between (disease) phenotypes ( phenomics ) in analogy to more frequent classification of gene relationships, protein-protein interactions, etc. similarity between phenotypes reflects gene relationships at the levels of 20 protein sequence and protein motifs protein-protein interactions (e.g. genes involved in the same multi-protein complex or biochemical pathway) functional annotation may be used to predict biological relations between genes and proteins (for functional annotation or disease gene prediction) MBC, Torino May 26th 2007

21 Phenomap van Driel et al., European Journal of Human Genetics 14: , phenotype map that describes the level of similarity between the majority of phenotypes establishes normalized similarity scores (between 0=unrelated and 1=identical/highly related) organized in a symmetrical score matrix: OMIM ID: Phenotype: # # # BREAST CANCER PROSTATE CANCER COLORECTAL CANCER, HEREDITARY NONPOLYPOSIS, TYPE 1; HNPCC1 PROTEIN C DEFICIENCY, CONGENITAL THROMBOTIC DISEASE DUE TO # [...] # [...] # [...] [...] # [...] # [...] # reasonable threshold for related phenotypes: 0.4 MBC, Torino May 26th 2007

22 Phenomap Construction (1) Problem: no systematic description of phenotypes in OMIM (hand-curated, different authors, etc.) Approach: Text-mining to classify >5000 phenotypes in OMIM extraction of phenotypic features of full text (TX) and clinical synopsis (CS) of phenotype entries search terms taken from anatomy (A) and disease (C) sections of the Medical Subject Headings (MeSH) vocabulary 22 hierarchical structure that groups MeSH terms into concepts useful for mining OMIM entries that use different terminology for the same concepts MBC, Torino May 26th 2007

23 Phenomap Construction (2) Construction of feature vectors one feature vector per OMIM phenotype entry each feature corresponds to a MeSH concept count the number of times the terms of a MeSH concept occur in the OMIM entry reflects the concept's relevance to the phenotype entry! the relevance is locally weighted (for each entry) according to how detailed it is (e.g. eye -> retina -> photoreceptors ) and how detailed the record itself is (shorter records have fewer occurrences) the significance of a feature/concept is also weighted according to the inverse document frequency measure: 23 non-significant concepts like syndrome or disease are excluded from the analysis global weight of a concept c: number of entries analyzed gwc = log2( number ) of entries that contain c MBC, Torino May 26th 2007

24 Phenomap Construction (3) Comparing OMIM phenotype entries: similarity quantified by comparing feature vectors (corrected according to local and global weight of MeSH concepts) similarity score s = cosine of the angle between the feature vectors a,b of two phenotype entries A and B: ai bi s a, b = 2 2 ai bi where ai and bi are the weighted frequencies of feature/concept i in entries A and B. 24 MBC, Torino May 26th 2007

25 Phenomap Evaluation evaluation on subset of 1653 phenotypes for which the causative gene and protein were known average of 10 randomized phenomaps as control for background signal results: 25 genes involved in similar phenotypes often have similar sequences allelic mutations of the same gene can cause similar phenotypes paralogs that have similar function can cause similar phenotypes genes that share functional protein domains (only partly similar sequence) similar phenotypes often share a protein-protein interaction or are related to the same pathway genes involved in similar phenotypes often share Gene Ontology (GO) annotations MBC, Torino May 26th 2007

26 References 26 OMIM: Hamosh et al., Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders, Nucleic Acids Research 30(1):52-55, website: Phenome Map / MimMiner: van Driel et al., A text-mining analysis of the human phenome, European Journal of Human Genetics 14: , website: MBC, Torino May 26th 2007

27 Exercises (1) What are the ten most similar phenotypes of Amyotrophic Lateral Sclerosis (ALS) in the phenomap? What can you find out about the most similar phenotype? 27 Of what OMIM category is its entry? Why? Is Hexosaminidase A Deficiency also related to ALS? What are the ID end class of its OMIM entry? What gene causes the disease? What is the most similar phenotype to Hexosaminidase A Deficiency? Why is it so similar? MBC, Torino May 26th 2007

28 Exercises (2) What is the most similar phenotype to Breast Cancer? Does it make sense to talk about most similar phenotypes? Is a phenomap score of so much better then ? 28 Is Breast Cancer also the most similar phenotype to that phenotype according the phenomap, i.e. does the reverse relationship hold? Remember how the phenomap is constructed and what assumptions are made Does the phenomap accurately reflect phenotype similarities? What is its strength/advantage? MBC, Torino May 26th 2007