General Technique
Pointillist is a set of four software programs for inferring the set of elements affected by a perturbation of a biological system based on multiple types of evidence. The inference is based on a weighted statistical method, in which each evidence type is given a distinct trustworthiness weight. The method is robust against missing and inconsistent data. Pointillist is written in Java, and is based on the ISB Java library. The software is available for download within a self-extracting installer. Pointillist requires the Java 2 Runtime Environment (JRE) version 1.4 or newer, or an equivalent JRE.
Purpose/use/application of the technique:
Pointillist consists of four programs:
- Data Manager
- Data Normalizer
- Significance Calculator
- Evidence-Weighted Inferer
All four programs read data files in which the columns are evidence types and the rows are elements of the system (e.g., genes or proteins). The Data Manager is used for data sub-selection and averaging. The Data Normalizer is (optionally) used to perform quantile normalization of microarray expression data. The Significance Calculator is used to obtain P-values for measurements based on a distribution of observations for negative control(s). Both parametric and nonparametric methods can be used to obtain the P-values. The Evidence-Weighted Inferer is used to combine the P-values for observations of different evidence types, and obtain a consensus P-value for each element of the system. The elements are classified into the "null hypothesis" (likely unaffected) set, and the set of likely affected elements, using an iterative procedure.
For more information about the Pointillist system, please consult the Pointillist User Manual.
Example(s) of projects at ISB that use this technique:
Pointillist has been used in the analysis of data for several studies here at the ISB, including a mouse prion study, a human prostate study, and studies of galactose uptake and peroxisome biogenesis in yeast.
Ongoing area of technology development:
We are developing a new version of Pointillist based on our evolving understanding of how best to integrate many types of high-throughput biological data.
|
home
