RAMP (Robust Accurate Modeling Protocol)

Class prediction is one of the most important goals in microarray gene expression data analysis. Successful class prediction is an important indication, not only of the importance of selected markers, but also of the robustness of the entire modeling process. Despite the fact that many publications report very high prediction performance, there has been a widespread lack of emphasis on robustness, reproducibility, and model generalization.

RAMP (Robust Accurate Modeling Protocol) is a predictive model construction package developed by Systems Analytics Inc. It is a software product for optimal and automatic construction of accurate, robust, and biologically relevant models for the prediction of control/treatment outcomes using microarray gene expression data.  

Feature Highlights:  

  • It is an integrated tool, which contains input, preprocessing, batch effect removal, model generation, model selection, output and future dataset prediction. Each of these components of the software is presented on one page.

  • It is easy to use! Users can generate and select robust and accurate predictive models by simply following the instructions and clicking the buttons! The software provides default options and parameters in every single step during the whole modeling process. Users can either follow the default options or modify them based their specific needs.

  • It can handle both one-color and two-color platform microarray data.  It can also be used for other types of high-throughput “omics” data such as proteomics and metabolomics data.

  • Batch effect is considered an important component of the modeling procedure. Users can view any batch effects in the training data with several visualization techniques, and determine whether or not to remove batch effect with the tools provided.

  • Multiple state-of-the-art algorithms are provided for each step of the model construction pipeline.  Users can simultaneously choose feature pooling method, feature selection method, and feature number subset.  They can also choose from a variety of classification algorithms and cross validation schemes.

  • Both feature pooling and feature selection are performed WITHIN the cross validation loop to prevent the class label information leakage.

  • A number of performance metrics are available, allowing the user to select and evaluate the models according to the criteria most relevant to their needs.

  • Users may also choose to use combinations of criteria to perform model selection.  This allows for an emphasis on the robustness and reproducibility of the predictive models.

  • The software provides an easy means of applying the selected models to the prediction of future unknown samples.

  • It is computationally fast!

 


RAMP Screenshots

Contact us for Free Trial


 
 
   
Copyright © 2006-2012 - Systems Analytics - Designed by Weboart