Skip to content
Browse files

Additions to guide pages

New guide page (detailing criteria for "sufficiently evaluated").

Also, added guideline against pooling data to achieve greater
statistical significant with case/control data.
  • Loading branch information...
1 parent d85cf33 commit 080da8ae61a29a1cb445958d35af797bf6762cd4 Madeleine Price Ball committed
View
4 public_html/guide_editing.php
@@ -21,11 +21,13 @@
h3. Which variants need editing?
+GET-Evidence uses collaborative editing to create and maintain our "variant interpretations":guide_reading_variant_reports. Which variants need editing?
+
A great place to start is the "public genomes":genomes that we have uploaded. To learn more about how to read these pages please see our "guide to reading genome reports":guide_reading_genome_reports.
If you open a genome report you'll find a tab listing "Insufficiently reviewed variants". These are variants which do not have enough scores and impact information filled in to allow us to automatically sort them. We sort these using "prioritization score":guide_prioritization_score, which prioritizes variants based on their presence in variant-specific lists (indicating there is published literature available), gene-specific lists (indicating potential for that gene to cause disease), and computational evidence (reflecting automatic prediction whether a variant may cause disease).
-"Scoring":guide_impact_score is the most important type of information to add to a variant interpretation. Once enough of these scores are added, a variant can be considered "sufficiently evaluated" by GET-Evidence.
+"Scoring":guide_impact_score is the most important type of information to add to a variant interpretation. Once enough of these scores are added, a variant can be considered "sufficiently evaluated":guide_sufficiently_evaluated by GET-Evidence.
You should also check out our "easy edit suggestions":guide_easy_edits to get started with easy ways to start improving GET-Evidence.
View
14 public_html/guide_impact_score.php
@@ -1,12 +1,14 @@
<?php
include "lib/setup.php";
-$gOut["title"] = "GET-Evidence: Variant Impact Score";
+$gOut["title"] = "GET-Evidence: Variant Impact Scores";
$gOut["content"] = $gTheTextile->textileThis (<<<EOF
-h1. Variant Impact Score
+h1. Variant Impact Scores
To facilitate automatic reporting of variants, we ask that users score variants in various categories. There are seven impact scores that are recorded, described below.
+For a variant to be considered "sufficiently evaluated" (and thus appear in a genome interpretation) some minimum number of these categories must be recorded. See our "guide to sufficiently vs. insufficiently variants":guide_sufficiently_evaluated for the specific requirements.
+
h2. Variant evidence vs. clinical importance
Of the seven impact scores, the first four reflect variant evidence (how "real" a variant effect is) while the last three reflect clinical importance (how severe and/or treatable the effect is). It is important to distinguish between these two sets of scores. Some variants may be predicted to have a very severe effect but have weak supporting evidence because the variant's rarity makes it hard to collect statistically significant data - such a variant would have high clinical importance scores but low evidence scores. On the other hand, some variants may have strong statistical significance but are common and have a weak impact on disease (for example, variants found through genome-wide association studies of common SNPs) - such variants would have high evidence scores but low clinical importance.
@@ -55,6 +57,14 @@
* Dominant hypothesis: Homozygous and heterozygous carriers are counted as case+, non-carriers are case-
* Counting chromosomes: Alleles rather than genotypes are counted, case+ is the number of chromosomes carrying the variant, case- is the number of chromosomes without it.
+h4. Do NOT combine data from different studies to increase statistical significance
+
+Publication bias is a serious issue that makes pooling data from multiple studies problematic. Pooled data from different studies fails to account for other studies where that particular variant was not observed, or was not reported upon (because it failed to have a strong assocation).
+
+Data from multiple studies should not be pooled if it strengthens the statistical significance of evidence supporting a variant. You may find cases where such pooling has been done in a variant evaluation. These cases may predate the creation of this guideline and should be corrected when found. You might, with discretion, combine data from other studies in a manner that weakens a hypothesis.
+
+In general, "meta-analysis" that combines and analyzes case/control data from different sources should be from peer-reviewed meta-analysis publications, not performed within GET-Evidence.
+
h3. Familial evidence
This category measures evidence based on the inheritance of this variant in families with the phenotype and how correlated the variant and phenotype are. For this category LOD (log-odds, using base10) scores are used:
View
4 public_html/guide_reading_genome_reports.php
@@ -37,9 +37,9 @@
h2. Insufficiently evaluated variants
-p. This tab contains a list of variants which haven't yet had a full interpretation performed within GET-Evidence. Each genome has thousands of variants affecting genes -- most of these do not yet have any interpretation associated with them.
+p. This tab contains a list of variants which haven't yet had a full interpretation performed within GET-Evidence. Each genome has thousands of variants affecting genes -- most of these do not yet have any interpretation associated with them. Once a variant has enough data recorded such that it is "considered sufficiently evaluated":guide_sufficiently_evaluated, it will appear instead in the "genome report" tab.
-p. To assist genome interpretation, GET-Evidence prioritizes these variants. When evaluating a new genome, a researcher might want to go to this tab and look for any prioritized variants that might be important to interpret. Once a variant interpretation is recorded it will be re-used by all other genomes that share that variant.
+p. To assist genome interpretation, GET-Evidence prioritizes insufficiently evaluated variants. When evaluating a new genome, a researcher might want to go to this tab and look for any prioritized variants that might be important to interpret. Once a variant interpretation is recorded it will be re-used by all other genomes that share that variant.
p(. !img/sample_genome_report_4.gif!
View
2 public_html/guide_reading_variant_reports.php
@@ -29,7 +29,7 @@
p(. !img/sample_variant_report_2.gif!
-p(. After the summary, the variant report contains a set of scores. These scores are generally entered by people who review the variant and help us automatically interpret the variant when producing reports.
+p(. After the summary, the variant report contains a set of "impact scores":guide_impact_scores. These scores are generally entered by people who review the variant and help us automatically interpret the variant when producing reports.
p(. The scores split into two sections: "variant evidence" and "clinical importance". These two sections reflect two very different aspects to understanding a genetic variant's impact -- how serious the effect is clinically, and how well-established the evidence is for that effect. For example, a variant might have been seen once in a family with a very severe genetic disease. The clinical importance of such a disease could be quite high. However, the strength of evidence supporting that variant as causing the disease could be extremely weak -- it was only seen that once, in one child, and not part of an extensive study comparing many patients and many controls.
View
31 public_html/guide_sufficiently_evaluated.php
@@ -0,0 +1,31 @@
+<?php
+
+include "lib/setup.php";
+$gOut["title"] = "GET-Evidence: Sufficiently vs. Insufficiently Evaluated Classification";
+$gOut["content"] = $gTheTextile->textileThis (<<<EOF
+h1. Sufficiently vs. Insufficiently Evaluated Classification
+
+Variants in GET-Evidence will not be listed in the main part of a genome report unless they are considered "sufficiently evaluated". Otherwise, they are listed in the "insufficiently evaluated" variants report, sorted by prioritization score.
+
+For a variant to be sufficiently evaluated, it needs to have "variant impact scores":guide_impact_score evaluated such that it can be automatically ranked in the genome report.
+
+h2. Criteria for "Sufficiently Evaluated"
+
+h3. Variant evidence scores
+
+At least one of either *"Case/control"* OR *"Familial"* scores must be recorded.
+
+Note that a score of zero is acceptable (i.e. if there is no significant evidence in the category). For example, a variant only once observed in a single child -- lacking both case/control and familial segregation data -- could score zero points in both categories.
+
+h3. Clinical impact scores
+
+For "benign" variants, no scores need to be recorded for this section.
+
+For other variants, both *"Severity"* AND *"Penetrance"* must be recorded.
+
+EOF
+);
+
+go();
+
+?>

0 comments on commit 080da8a

Please sign in to comment.
Something went wrong with that request. Please try again.