Binning clones by hybridization with complex probes: statistical refinement of an inner product mapping method

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Top-down molecular strategies, utilizing long-range genomic information, are increasingly vital for genomics research.
• Inner Product Mapping (IPM) is a key top-down strategy that uses radiation hybrid (RH)/YAC and RH/STS overlap data for clone localization.
• Existing IPM methods are sensitive to noise in molecular data, necessitating improved statistical approaches.

Purpose of the Study:

• To develop advanced statistical models for Inner Product Mapping (IPM) to enhance clone localization accuracy.
• To implement these models within a hierarchical Bayes framework for robust genomic analysis.
• To refine the existing chromosome 11 map using the developed statistical models.

Main Methods:

• Development of two statistical models for IPM: a parametric logit model and a nonparametric order-restricted model.
• Implementation of these models within a hierarchical Bayes framework.
• Application of the models to refine the chromosome 11 map using RH/YAC and RH/STS overlap data.

Main Results:

• The developed statistical models significantly improve the effectiveness of IPM by accounting for noise in molecular data.
• Refinement of the chromosome 11 map resulted in enhanced successful localization of clones.
• Increased confidence in the localization of clones on the chromosome 11 map was achieved.

Conclusions:

• Statistical modeling is crucial for optimizing top-down genomic strategies like IPM, especially when dealing with noisy molecular data.
• The hierarchical Bayes framework provides a robust platform for implementing these advanced statistical models.
• This work successfully refines the chromosome 11 map, demonstrating the practical utility and improved accuracy of the developed methods.