RESEARCH PLAN

Principal Investigator/Program Director Williams, Robert W.

The Neurogenetics Tool Box

Our plan in this section of the application is to develop unique World Wide Web services that will make it possible for neurogeneticists to rapidly identify and map genes and quantitative trait loci, particularly those related to brain structure and behavior. The Neurogenetics Tool Box (NTB), which our group at the Roswell Park Cancer Institute will design, test, and distribute on the web, will be tightly integrated with the MBL and NeuroCartographer Projects. Synergy between these components of the program project will allow neuroscientists and geneticists to explore the complex genetics of CNS architecture and to map QTLs much more efficiently than is now possible. The NTB will consist of three software modules and complete online documentation and tutorials.

1. The NTB biostatistical evaluation module

An essential prelude to gene mapping is to analyze the statistical properties of the trait, in particular, its normality, its covariance with other phenotypes, and of course, its heritability. We will construct a web interface using Python scripts, CGIs, and custom server software that will allow neuroscientists to submit quantitative trait data from any sample population. In response, this NTB module will rapidly return (1) key descriptive statistics, (2) results of tests of normality and homoscedasticity, (3) correlation matrices between submitted traits and other selected traits (e.g., brain weight, body weight, age, and sex), (4) estimates of narrow and broad-sense heritability, and, when appropriate, (5) estimates of the minimum number of effective genes controlling CNS traits. This system will include modules which clients will use to perform transformations of trait values to improve normality, to control for nongenetic variance (e.g., age), to test for outliers, and to test for uniformity of variance across complex data sets. Clients will be able to rapidly evaluate the suitability of their phenotypes for gene mapping without becoming expert statisticians.

2. The NTB gene and QTL mapping module

The web interface of this system will allow rapid QTL mapping of CNS traits for major recombinant inbred (RI) sets, for shared intercrosses and backcrosses, and particularly for a very large advanced intercross population (the G₁₀) described in the Genotyping and Mouse Colony Core. The NTB will include curated data for marker genotypes and map distances for all crosses and RI strains in the Mouse Brain Library. Using the resources developed by this program project, neurogeneticists will be able to generate phenotypes, then feed the data directly into the NTBs biostatistics and gene mapping modules. Furthermore, the NTB will include an extensive legacy database on behavior and neuropharmacology of RI strains. (The core of this database will be provided by John Belknap, a member of the External Advisory Committee.) It will therefore be feasible to explore the genetic basis of differences in a broad range of CNS traits.

The NTB gene mapping module will initially be based on algorithms already successfully implemented in the programs Map Manager QT and Map Manager QTX (Manly and Olson 1999). The NTB will perform single-locus association tests, simple interval mapping, and composite interval mapping and will also search for epistatic interactions between QTLs. The mapping module will be implemented in Python and C++, with Python used for interaction with web forms and data formatting and C++ for the analysis routines.

For functions in which processing speed is not limiting, results will be returned to clients immediately in the form of web pages; for lengthy calculations, data sets and figures will be returned by email. This system will include empirical significance thresholds established by permutation tests or by a new method that promises to be much faster (H-P. Piepho, personal communication). Our NTB mapping program will also allow for weighted linear regression for those data sets in which the variance of phenotype values is available, and it will provide nonparametric mapping methods for traits with distributions far from normal.

The web interface of the mapping module will also allow clients to submit private data sets (trait values, marker genotypes, and map distances) for QTL analysis in crosses that are not curated by the NTB. Clients will have access to the same range of analysis options as described above except that correlation with other traits will necessarily be limited to those traits submitted with the population marker data. A high level of confidentiality will be guaranteed to clients using this automatic gene mapping system. As part of this system, we will assemble and maintain an extensive database on phenotypes and genotypes for the major recombinant inbred strains (BXD, AXB-BXA, CXB, BXH, AKXD, etc).

A powerful and unique feature of the gene mapping module is that it will allow investigators to search for genetic correlations among large numbers of traits and QTLs in our NTB archive. The NTB databases will become progressively more powerful as more traits are mapped on the same crosses. We anticipate that networks of gene interactions and gene pleiotropy will be exposed.

3. NTB data archiving and export modules

4. NTB documentation and tutorials

Next Topic

• Problems and issues addressed.