Project

General

Profile

Actions

Wiki » History » Revision 71

« Previous | Revision 71/88 (diff) | Next »
Jason Bobe, 03/11/2010 08:08 PM
ok I think I've got lod down. check out: http://www.ncbi.nlm.nih.gov/bookshelf/br.fcgi?book=hmg&part=A1378


= Genotype + Environment = Trait Evidence (GET-E) Database =

The '''GET-E''' project was started by Abraham Rosenbaum, Joe Thakuria, Xiaodi Wu and Alexander Wait Zaranek to help the [http://personalgenomes.org Personal Genome Project], the clinical genetics community, and, the general public interpret individual genomes. The database is closely integrated with [wiki:Trait-o-matic].

You can find the PGP production server at: * http://evidence.personalgenomes.org

A development sandbox can be found at: * http://evidence-dev.personalgenomes.org

Many contributors helped make '''GET-E''' possible. You can find specific contributions at: * http://evidence.personalgenomes.org/editors

Please help the project by editing the database, writing software tools that use the database (eg. Trait-o-matic/AETAS) or by making suggestions on the wiki!

Field Descriptions and Editing Guidelines === Short Summary === * free text providing a one line summary of clinical action to be undertaken given this variant (possibly modified by known phenotypes).
=== Variant Evidence === 
Each of these fields is a sliding scale of 0-5 where 0 represents no evidence and 5 strong evidence within the given category. Where the data is not entered the default is N/A. When Genetests and/or the literature correlate this variant with a number of phenotypes the impact should be reported for the most extreme condition
a. Computational: One star for each consistent prediction and one star subtracted for conflicting results from:
i. evolutionary conservation (minimum of three species)
i. presence in active domain
i. SIFT
i. !PolyPhen
i. NBLOSUM >= 2
i. GVGD
i. Other variants in this gene cause similar disease
i. etc...
a. Functional: One star for each experiment supporting the result, and penalize one star for conflicting results from:
i. enzyme extracts
i. cell lines
i. animal models
i. etc...
a. Case/Control: A combination of odds ratio and significance test (currently using Fisher's Exact Test). For protective alleles, use inverse OR = 1 / OR. Do not count related individuals, count probands -- ie. one per family
i. 0 stars if no higher ranking is allowed
i. 1 star if OR>1 and significance <= 0.1
i. 2 stars if OR>=1.5 and significance <= 0.05
i. 3 stars if OR>=2 and significance <= 0.025
i. 4 stars if OR>=3 and significance <= 0.01
i. 5 stars if OR>=5 and significance <= 0.0001
a. Familial:
i. -1 stars if best LOD < 0.5 and LOD < -2 for theta = 0.1
i. 0 stars for no familial information
i. 1 star for best LOD >= 0.5
i. 2 stars for best LOD >= 1
i. 3 stars for best LOD >= 1.5
i. 4 stars for best LOD >= 3
i. 5 stars for best LOD >= 5
=== Disease Importance ===
As with the Variant Quality section, each of these fields is a sliding scale of 0-5. These two categories are not applicable for benign and protective variants.
a. Severity: downgraded according to disease penetrance (eg. Crohn's disease would be moderate or severe, but "increased susceptibility" could only mean that the chances are increased by ~.15% and so would be called mild). For pharmacogenetic variants this is the severity of using the wrong drug/dosage.
i. 0 stars for benign
i. 1 star for very low expectation of having symptoms for this genotype, very low penetrance (eg. Susceptibility to crohn's with OR = 4.5, causing a ~.2% risk of Crohn's)
i. 2 stars for mild effect on quality of life or unlikely to be symptomatic (Cystinuria)
i. 3 stars for moderate effect on quality of life (eg. Familial mediterranean fever)
i. 4 stars for severe effect: causes serious disability or reduces life expectancy (eg. Sickle-cell, Stargardt's disease)
i. 5 stars for very severe effect: early lethal (eg. Familial adenomatous polypopsis, adrenoleukodystrophy)
a. Treatability:
i. 0 stars for no clinical evidence supporting intervention (eg. PAF acetylhydrolase deficiency)
i. 1 star for uncurable: treatment only to alleviate symptoms (eg. adrenoleukodystrophy)
i. 2 stars for potentially treatable: Treatment is in development or controversial
i. 3 stars for treatable but a significant fraction do not require treatment (Cystinuria)
i. 4 stars for treatable: Standard treatment reduces the amount of mortality/morbidity but does not eliminate it (eg. Sickle-cell disease)
i. 5 stars for extremely treatable: Well-established treatment essentially eliminates the effect of the disease (eg. PKU)
=== Impact ===
This is filled in as one of the following:
a. Pathogenic: Causative for disease.
a. Benign: No clinically significant phenotype.
a. Protective: Protective from disease.
a. Pharmacogenetic: Clinically significant phenotype in the presence of a pharmacological agent.
a. NR: Not reviewed.

In addition to these categories, up to two qualifiers will be added automatically according to the variant quality scores. The first, "evidence", is based on the four "Variant Evidence" categories. The second, "importance", is based on the two "Clinical Importance" categories (not applicable for benign and protective categories).

An example for "pathogenic" follows: || - ||'''low evidence'''||'''moderate evidence'''||'''high evidence''' ||'''low importance'''|| uncertain pathogenic || likely pathogenic || pathogenic ||'''moderate importance'''|| uncertain important pathogenic || likely important pathogenic || important pathogenic ||'''high importance'''|| uncertain very important pathogenic || likely very important pathogenic || very important pathogenic

Criteria for evidence:
a. For no qualifier (ie, well-established): (1) at least 4 stars in either "Case/control evidence" or "Familial evidence" and (2) at least eight stars total
a. For "likely": (1) at least 3 stars in either "Case/control evidence" or "Familial evidence" and (2) at least five stars total
a. Otherwise "uncertain"

Criteria for importance (not used for benign and protective):
a. For "very important": (1) at least 3 stars in severity and at least 4 stars in treatability, or (2) at least 4 stars in severity
a. For "important": (1) at least 2 stars in severity and at least 3 stars in treatability, or (2) at least 3 stars in either severity
a. Otherwise no qualifier

When Genetests and/or the literature give multiple phenotypes for a variant, the impact should be reported for the most extreme condition and annotated as such in the ''short summary''. In the future we hope to list different phenotypes separately.

=== Inheritance Pattern === 
a. Dominant
a. Recessive
a. Other: A defined form of inheritance that is neither dominant or recessive (e.g. modifier, co-dominant, incomplete penetrance). These variants will get prioritized due to their potential phenotypic affect should the additional factors be present.
a. Undefined: Undefined in the literature.
a. Unknown: The default value for all variants and all variants without literature.
=== Summary of published research, and additional commentary ===
  • Free text providing a comprehensive review of the variant including youngest age of onset, oldest age of onset and oldest asymptomatic individual.
=== Allele Frequency ===
  • Automatically generated frequencies for the !HapMap Project, 1000 Genomes Project and the weighted average of the two. Sub-population frequencies and homozygous/heterozygous break-downs are currently not shown.
=== Publications ===
  • User-entered publications and user entered synopses of the relevant details from the publication. A short summary is also entered for each paper (as above).
=== Genomes ===
  • Automatically generated list of all genomes from Trait-o-matic in which this variant is seen. Additionally, specific actions taken by each individual harboring the variant can be reported.
=== Other External References === 
These are automatically generated by web "robots":
a. dbSNP: The link to the dbSNP entry for this nucleotide position (for this nucleotide change?)
a. Genetests: Link to all clinical laboratories performing diagnostic tests on this gene and the list of diseases that the labs are looking for.
a. Web Search Results: A web search for this exact variant using the Yahoo search engine
=== Edit History ===
  • Automatically generated history of all page edits with the contributor's name.
To Do * [wiki:GET-Evidence/23andWe] Suggestions
  • your suggestion here.

Updated by Jason Bobe about 14 years ago · 71 revisions