Computational Biology Research Center[CBRC]

MISSION

Development of "POODLE", a disorder prediction program

MEMBERS

Tamotsu NOGUCHI : CBRC

Shuichi HIROSE : Pharma Design, Inc.

Kana SHIMIZU : CBRC, Waseda University

Satoru KANAI : Pharma Design, Inc.

Interview : 26th Apr. 2007

Some researchers at CBRC are tackling and have achieved results in relation to the issue of "protein structure prediction". A team led by Deputy Director Noguchi is one team involved in this research. The prediction of disorder regions (regions with a fixed 3D structure in a　protein) that this team has developed is the first step in protein structure prediction, and the high level accuracy of this new prediction method has been acknowledged worldwide.

Protein structure prediction is one of the main research fields in bioinformatics. This is a research to predict the tertiary structure from the primary structure (amino acid sequence) of a protein accurately. Determination of protein structures contributes a great deal to research in other fields that are in the public interest, such as drug design. Therefore, researchers all over the globe are continuing to tackle the challenge of improving the accuracy of protein structure prediction.

Protein structure prediction is an intricate research theme (because the amino acid sequence pattern in proteins is huge, and the structure of each one is different). The degree of difficulty for research in this filed is also raised by the fact that there are many facets of this research for which a physical basis has not yet been clarified, in relation to the conditions under which the structure is steady.

I want to achieve good results in my research, and display them with pride

Noguchi was worried about how to win the category of "Order-Disorder Regions prediction" of CASP7 (7th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction) held in May 2006. He was concerned about a suitable way for development of a prediction program capable of winning the category of CASP7. CASP was like Noguchi's "home ground". Noguchi, who has carried out research in the field of protein structure prediction for over 20 years, understands just how difficult the targets prepared for CASP are, and also the glory attained by doing well in the contest.

Looking back on this period, Noguchi reflects, "CASP5 was the first time that a disorder region prediction category was established in CASP. At that time, I was interested in competing in other categories, so I didn't take on the disorder region prediction targets. I first tackled them it at CASP6, and achieved an average result, in part because I didn't prepare properly. From CASP6, the number of researchers participating in this category has increased. At CASP7, my desire to achieve good results, and to display the results of my research with pride, became stronger."

"I want to achieve one of the top prizes."
What kind of contest is CASP, one that elicits such motivation?

There was no time to relax

The contest Noguchi participated in, CASP, is a contest of the protein structure prediction technique held at a global level once every two years. Researchers from all over the world compete with each other at the contest for their skill and the prediction accuracy of their respective methods. The competition is decided by assessing the submitted prediction results of targets provided by CASP. The researchers compete for the prediction accuracy of the targets provided by CASP Organizing Committee, which has already been finished the analysis of the structure, but unpublished the result; in other words, it is a blind test.
The primary structure of the protein was released, and participants were to conduct predictions for a category of tertiary structure, domain boundary, or like Noguchi, a disorder region in the contest, and submitted their results. At CASP7, 100 targets for prediction were provided for a category. Participants were to keep predicting the structure and to submitting the result in a limited time (two to three weeks) during the period allotted (for CASP7, this was the three month period between May and August 2006). " It is really head for the CASP season. We would have predicted the disorder regions for two or three targets, and closely examine whether the predicted results would be correct or not every day. If a result is not a good, we will return to the first step of the prediction. There wasn't really any time to rest."