QUICK FACTS
Programmers in the Statistical Genetics section of the Department of Biostatistics at the University of Alabama at Birmingham are using the JMSL Library for a High Dimensional Biology statistical application. They are analyzing micro-array data using statistical methods from the JMSL Library.

THE PROBLEM
Biostatistics is the field of statistical methods related to biological research in areas such as public health, medicine, agriculture, dentistry and nursing. At the University of Alabama at Birmingham, the Department of Biostatistics focuses on the application of existing statistical methods to studies in these agricultural and health related fields and the development of new statistical techniques.

Section Lead, Dr. David B. Allison was looking for a numerical library to aid in the development of an application for the analysis of HDB data in Statistical Genetics research. The term “HDB” stands for High Dimensional Biology, which is a fairly recent term coined to describe the “omic” sciences that are becoming quite popular areas of research such as genomics, proteomics, etc. Allison and colleagues wanted to concentrate their efforts on writing proprietary algorithms for the application and implementing the novel analysis methods developed by the Section on Statistical Genetics in the School of Public Health into their application HDBStat!.

Due to the fact that they have a small team of programmers, they were looking for an existing, yet highly reliable, solution so they would not have to “reinvent the wheel.” They searched for a sophisticated numerical library that performed the analysis they needed in a quick and efficient way.

THE SOLUTION
A few of the specific traits they were looking for in a numerical library included summary statistics, t-tests, non-linear optimization, random number generation, and PDF’s and CDF’s for common and uncommon distributions. This functionality needed to be in Java to integrate into the application the Biostatistics group was developing. They found the building blocks they required in the JMSL Library for Java Applications from Visual Numerics.

The JMSL Library is a complete collection of mathematical, statistical and charting classes, written in 100% Java, which provides the ability to develop network-centric, cost effective data analysis applications. It is the only solution to combine integrated charting with the reliable mathematical and statistical functionality of the industry-leading IMSL™ Library algorithms.

Allison and colleagues found that the JMSL Library implemented the features necessary to be tremendously useful to his group. Visual Numerics has made it possible to perform simple to complex data analysis on the Java platform by developing a unique set of custom language extensions in the JMSL Library. These extensions provide the basis for optimized analytics on one of the most flexible and collaborative platforms available. In addition, the JMSL Library provides a strong Java API for its numerical functions.

Furthermore, Visual Numerics is committed to work with Allison’s team in the Department of Biostatistics to incorporate additional features they would need for future applications. This type of commitment and close working relationship proved very valuable to UAB. All of these factors and Visual Numerics 30-year track record in numerical analysis led Allison’s group to select the JMSL Library for their work.

As a result, the UAB Statistical Genetics group is using the JMSL Library in their HDBStat! application that is used to analyze micro-array data by using known statistical methods as well as novel ones developed within the Section on Statistical Genetics.

RETURN ON INVESTMENT
“Utilizing the JMSL Library redirects the precious time of our software developers that would otherwise be spent working on accurate and reliable algorithms already present in the Library. Now their focus can be directed at implementing and refining new ideas specific to our work,” explains Allison.

Allison’s level of satisfaction with the JMSL Library is very high. The faster development times combined with shorter development cycles have made their analysis and research in the area Biostatistics much more focused and cost effective. As for the technical support team at Visual Numerics, “I couldn’t be happier,” says Jelai Wang, lead developer of HDBStat!, “The support has been professional and all of my requests and questions have been handled in a very timely manner.”