UAB Division of Informatics to manage medical research data

Which brings us to Jonas Almeida. Almeida is a recent addition to the University of Alabama at Birmingham medical center, recruited away from the prestigious University of Texas-MD Anderson Cancer Center in January 2011 to launch a newly formed Division of Informatics in the med school's pathology department.

At MD Anderson as a professor of bioinformatics, Almeida was deeply involved in the Cancer Genome Atlas, a shared site for cancer data started with the help of National Cancer Institute funding in 2006. Its utility was seen just last week in a TCGA study of gene mutations in colon and rectal cancers.

Researchers used the shared data to show that the patterns in colon and rectal cancers were the same, regardless of where they came from. This means the two are actually a single type of cancer. The discovery is considered an important step in understanding the foundations of that disease.

An effort like the Atlas, Almeida said, takes advantage of "the cloud" — a web-based set of programs and standardized data formats that allows researchers to build interactive masses of data. Almeida likened this use of the cloud to the popular pastime FarmVille, a farming simulation social network game.

Another approach in informatics is seen in a paper Almeida and colleagues published Friday about a new app they created for image bioinformatics. This app, called ImageJS, is like one version of Angry Birds, because it operates inside a Google Chrome web browser.

ImageJS started with a common pathology problem — a patient has had a brain biopsy, and a slice of the brain has been put on a slide and stained with one dye to color the nuclei of cells and with another dye to create a different color in nuclei that are replicating.

Normally a pathologist will look at an image of the stained cells and through individual skill and possible help from image processing software, figure how much replication is going on. A high amount means cancer.

"Pathologists face two major problems that have been insoluble for many years," Almeida said in a telephone interview from Germany.

First, they would run into difficulties about patient privacy if they had to send the images off to some other computer site to be analyzed, and any software they might want to put on their own hospital computers would create a security question.

Second, if they wished to make any changes to the algorithms used to analyze images, they would need to get the software rewritten by its producer, and then go through the long and tedious process of getting that software cleared by the information technology folks.

ImageJS, a simple and fast program written in JavaScript, sidesteps both of those problems.

It lets the pathologist work at his or her own computer but avoids adding any new programming files to the computer. The ImageJS app does this by operating in a Google Chrome browser which acts as a "sandbox where code is executed without direct access to the machine's file system," Almeida, pathologist Dr. James Hackney and four other division of informatics colleagues wrote in their Journal of Pathology Informatics paper.

Free code

Furthermore, the code is freely available from hosting services such as GitHub and Google Code. Users will be able to write a few lines of code to slightly alter the image algorithms, something akin to an Angry Birds player being able to change the colors of the birds, Almeida said.

While ImageJS addresses a pathology problem, it also doubles as an experiment in informatics. Almeida said he hopes to see it lead to interactions where researchers use and modify the app's simple architecture to build a collaborative computational ecosystem.

[See Almedia's YouTube.com channel.]

At UAB, the Portuguese-born Almeida has hired five faculty and now has about 15 researchers and staff in the informatics division.

UAB pathology chairman Dr. Kevin Roth was the key to luring Almeida to UAB.

"He offered me something that is special — a strategic vision about the future," Almeida said. "He sees biostatistics and bioinformatics to be part of the tools of select medical centers."

Almeida gave another example of the potential use of bioinformatics for clinicians — shared cancer databases in the cloud where oncologists can find answers to questions that could improve patient care.

Such questions, Almeida said, include, "What patients have been seen before that were like the one that I'm seeing? What happened to those patients? What drugs did they respond to?"

Join the conversation by clicking to comment or email Hansen at jhansen@bhamnews.com.