The variation-event class can be confusing. In part, this is because a variation-event isn't referred to by just a single string in the text. Rather it is an evoked entity, meaning that it emerges from the relationship between the variation-entity subcomponents (type, location, and states). When the variation subtypes are combined to describe a single variation, then this is a variation event.
In this document I will present a formal definition for our concept of Variation-Event.
Each of the three basic variation-entity types is described in detail in other documents. Below is a brief review of how the variation-entity class relates to the project as a whole, as well as a brief introduction to the types of variation sub-entities.
The goal of the project is to define the following relationship between tagged entities:
A Genomic Element undergoes a Variation Event which is correlated with a Malignancy
This relationship captures an easily understandable progression: namely, that there exists a correlation between some genomic element, a variation in the genomic element, and a malignancy. Based upon our definition of the Gene-Entity class (please read that description if you have not done so already) we know that the genomic element is a gene, protein or transcript which has a certain function (the conceptual notion of a gene as described). Therefore, a malignancy is the result of an alteration to some genomic element that changes (or "breaks") the normal function. In the literature, this cause-effect relationship is not always clear. More frequently, the literature describes a correlation or association between a genomic element and a malignancy (sufficient for our study) which can be further tested to isolate the precise mechanism.
Therefore, a Variation-Event is the "action" of the genomic element's change. This "event" is described by specific details, which are interrelated for any specific variation: Event = Location + Type + Specific Change:
As mentioned, the location of the variation event is the crux of the variation-entity class. However, that does not mean that it will always be given. The other aspects of the variation-entity describe what is going on at the variation-location.
This describes the detailed change that has occurred at the location in question. This is the most specific description of the alteration possible. Combined, variation-location and variation-specific-change provide all of the details necessary to account for the aberrant phenotype. However, these two classes may not always be present together.
All specific alterations can be grouped into more general alteration classes. Examples of such classes are point mutation, translocation, and deletion. Because of the idiosyncratic descriptions of variation, we will not always be given the complete information. There will be times when the variation-specific-change is NOT given but a more general description (namely variation-type) is used instead. Occasionally, knowing the variation specific change will illuminate the variation type (i.e "A35T" and "point mutation"), but this is not always the case (e.g., "deletion" can be hemizygous or homozygous).
Most of the entities we have dealt with up till now have mapped fairly closely to the text that referred to them.* But a Variation-Event is typically what we call an "evoked entity": instead of understanding from a single reference, the reader assembles an understanding of it from a number of simpler entities referred to in the text and the relations between them, either explicit or implicit. These component references may be scattered across a broad stretch of text. We do not tag the text describing a Variation-Event; we tag only the component references. We use the Variation-Event tag only for compact references to an evoked entity described elsewhere in the text.
[* We have not distinguished consistently between entities and references to entities, but the distinction is important. A string of text, such as "Ki-ras" or "deletion at codon 57", refers to a thing (allowing a very broad definition of "thing"), or an event or to a class of things or events. Strictly speaking, it is the "things" that are the entities; the chunks of text that we tag are references to them.]
Let us look at a piece of text:
"The gene MYCN undergoes a point mutation at base pair 135,245 changing A to T."
From this example, it is clear that we are referring to a single Variation-Event with the following subtypes: "point mutation" (type), "base pair 135,245" (location), "A" (state, initial), and "T" (state altered).
Variation-Event:
| Type | point mutation |
| Location | base pair 135,245 |
| State -- initial | A |
| State -- altered | T |
The text clearly refers to a variation-event, made up of these subcomponents. This is what we mean by an evoked entity. Therefore, a variation event is defined by its subcomponents. It exists as the evoked entity of a specific combination of subcomponents. It does not exist in the absence of clearly stated subcomponents. Identifying subcomponents in this manner defines a variation-event.
Imagine later in the text we encounter...
"This point mutation is correlated with neuroblastoma in 80% of patients observed."
From this example we can infer that "This point mutation" refers to the variation-event defined by the aforementioned subcomponents. "This point mutation" is being used as a short-hand notation to refer to that specific variation event. Therefore, "This point mutation" can be tagged as a variation-event because it refers precisely to a single described variation event.
In conclusion:
Written by Scott Winters and Mark Mandel. Last updated 2004-05-05