{no title found}

Quality and confidence are distinct concepts: quality refers to the value of a source or of an annotation, whereas confidence refers to the level of certainty that an assertion is correct.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Different strategies can be used to define the overall confidence level: for instance, it could be defined based on the confidence level of the most trusted evidence, from the set of evidence lines supporting (not contradicting) the assertion; or, it could be defined based on the number of evidence lines supporting the assertion (e.g., 10 congruent low-confidence evidence lines could allow to have an overall high confidence in an assertion, even if contradicted by one other low-confidence evidence). It is the responsibility of annotation teams to define and communicate the parameters, pertinent to their field, allowing to produce overall confidence levels.

Subclasses are used to define whether there is a single evidence, or multiple evidence lines, in support of an assertion, and whether they are congruent or conflicting.

The assumption here is that assertions supported by a single evidence are less trustworthy than assertions supported by multiple evidence lines.

An assertion can have several related evidence lines, some supporting the assertion, and some rejecting the assertion (e.g., using the NOT qualifier in GO annotations).

We distinguish two levels of conflict: weak conflict between evidence lines, e.g., a low confidence evidence negating an assertion, versus several high confidence evidence lines supporting the assertion; and strong conflict between evidence lines, e.g., two high confidence evidence lines contradicting each other. Note that strongly conflicting evidence lines are not used to produce confidence information statements associated to a confidence level (low, medium, high): in that case, the supported assertion is believed to need further validation, as evidence lines yield different conclusions of similar confidence.

We distinguish two levels of conflict: weak conflict between evidence lines, e.g., a low confidence evidence negating an assertion, versus several high confidence evidence lines supporting the assertion; and strong conflict between evidence lines, e.g., two high confidence evidence lines contradicting each other.

When results are retracted, associated annotations are often removed. Consequently, end-users might not be aware that the result was annotated. Annotating an assertion with this confidence term allows to keep the information available. Such assertions should not interfer with aggregation of multiple evidence lines to generate a confidence level from the branch CIO:0000002 "confidence from multiple evidence lines". This is different from negative annotations, used to negate the interpretation of an annotation. It is also different from conflicting multiple evidence lines, where each single evidence has not been directly invalidated. Rejection is a stronger assertion about an annotation source than conflict, capturing that this evidence should no longer be used.

The assumption here is that assertions supported by evidence lines of different experimental or computational types are more trustworthy than assertions supported by evidence lines of a same experimental or computational type.

The assumption here is that assertions supported by a single evidence are less trustworthy than assertions supported by multiple evidence lines.

The assumption here is that assertions supported by a single evidence are less trustworthy than assertions supported by multiple evidence lines.