Thoughts on Representing QTL Data in MGD

What are QTLs?

Representation Issues

Questions Arising From Trying to Code QTL Papers

Conceptual Model

QTL Query Examples

Literature

What Are QTLs?

Quantitative Trait Loci (QTL) are hypotheses that specific chromosomal regions contain genes that make a significant contribution to the expression of a complex trait.

QTLs are generally identified by comparing the linkage (degree of covariation) of polymorphic molecular markers and phenotypic trait measurements.

The ultimate goal of complex trait dissection is to identify the actual genes involved in the trait and to understand the cellular roles and functions of these genes.

The accuracy and precision of locating QTLs depends, in part, on the density of the linkage map created. The higher the density of the map, the more precise the location of the putative QTL. When QTLs can be mapped to a relatively small chromosomal region or regions other methods, such as positional cloning, can be used effectively to isolate specific genes. Unfortunately, the denser the map, the more likely that false positive QTLs will be detected.

Most, but not all, complex traits are conditioned by more than one locus. QTLs often interact in complex ways and their expression can also be influenced by non-genetic factors.

Because QTLs are hypotheses, they are subject to reinterpretation and revision. Because the location of QTLs are provisional their nomenclature is likely to be fluid and temporary.

Some of the terms associated with QTLs (phenotype, trait, etc.) are semantically "fuzzy." The definitions I used in developing a conceptual model of QTL data can be found HERE .

A useful description of a QTL includes at least the following elements (with indications of the certainty associated with each point):

• What traits were used to define a complex phenotype
• What mice strains/progenitors were used in a study
• The type and size of the mapping population
• Methods used for QTL mapping
• Chromosomal location(s) of the QTL
• Candidate genes in the QTL region
• Magnitude of effect of the QTL on phenotype (% phenotypic/genetic variance accounted for)
• Model(s) of gene action (additive, dominance, etc.)
• Epistatic interactions, if any
• Comparison of QTL information to other comparable studies in mouse
• Comparison of QTL information to studies of the genetic basis of similar complex trait phenotypes in other organisms (especially, human)

No Time to Browse! I'm on the Quick Tour!

Thinking about these broader issues in more detail:

1. How the phenotype of a complex trait is defined can affect how easy it is to identify and map QTLs.
2. How much of the raw data should be represented in the database?
3. There is a fair amount of uncertainty associated with QTLs, including:
   • uncertainty in whether a QTL is really associated with the complex trait in question,
   • uncertainty as to the chromosomal location of the QTL,
   • uncertainty in whether the QTL contains one or more linked genetic factors,
   • uncertainty in nomenclature,
   • uncertainty as to which genes/alleles in the mapped QTL region are likely to be directly related to the complex trait in question.
4. The data model for QTLs should provide representations for the following:
   • multiple crosses
   • multiple analysis methods
   • versioning of QTL experiment data as previous findings are revisited and revised
   • model of QTL action if proposed
   • uncertainty in the data as outlined above in #3
   • # animals and/or strains used
   • information on the phenotype of complex trait
   • information on the strains used and their phenotypes
   • method(s) used to create linkage map(s)
   • intermarker range
   • information on statistical significance of QTL association with complex trait
   • information on chromsosomal location - including links to graphical representations of chromosomes and more detailed marker maps in those regions
   • heretibility
   • % total phenotypic variation and % genetic variation a QTL accounts for
5. Access points to the Complex Trait/QTL data will likely be through
   • Phenotype
   • Strain
   • Marker
   • Candidate loci?
   • Chromosome
   • Author(s)

• For induced phenotypes, report experimental conditions (e.g., dosage and delivery mechanism)?
• Report non-significant LOD scores, etc. when presented in paper?
• What data format to allow for representing chromosomal location data?
• How to report data for multiple passes at QTL analysis (e.g., first including all phenotypes and then including only the extremes)
• Reporting % total phenotypic variation VS % genetic variation
• Report preliminary QTLs formally or as notes (see J# 28399)
• How to report results when two crosses show different results (see J# 37769)
• Crosses reported to MAP QTLs and to measure their effects -- how to report the second case? (see J#37769)
• What to do when author doesn't propose QTL name?
• How to report data when a chromosome has sig LOD scores in two different areas and two different QTL models are proposed.
• Do we want to provide links to the persons associated with a study like they do in the plant genome community?

• What QTLs are hypothesized to be associated with susceptibility to lung tumors in mice?
  • What % of phenotypic variation is associated with each QTL?
  • What is the chromosomal location for each QTL?
  • Show me all probes associated with these QTLs
  • What are the significance values for each QTL?
• How many QTLs have been reported where less than 100 animals were used in the analysis?
• List all QTL studies (by J #) where sex, age, or paternity have been reported to play a signficant role in phenotypic variation.
• Are any of the QTLs identified in my cross similar in predicted chromosomal location to QTLs demonstrated from another cross?
• Do the QTLs in mouse identified for epilepsy correspond to any QTLs for epilepsy identified in syntenic regions in human (or some other species)?
• Have any other phenotypic traits been mapped to chromosome segment that my QTL appears to fall into?

Check It Out!

BACK TO THE QTL RESOURCE HOME PAGE