Nutritious Rice For The World [complete], рассчет популяции нового, более питательного риса Налаштування

[Rilian]

Nutritious Rice for the World

проект стартовал 12 мая 2008

Полезные ссылки:

• Интервью с Dr. Ram Samudrala, координатором проекта

Как присоединиться читайте в главном топике World Community Grid

Изучается строение протеинов риса, чтобы потом, не используя генeтические модификации, а просто за счет скрещивания, выводить новые виды. :flowers:

Mission
The objective of this project is to predict the structure of proteins of major strains of rice. The intent is to help farmers breed better rice strains with higher crop yields, promote greater disease and pest resistance, and utilize a full range of bioavailable nutrients that can benefit people around the world, especially in regions where hunger is a critical concern.

Determining the structure of proteins is an extremely difficult and expensive process. However, it is possible to computationally predict a protein's structure from its corresponding DNA sequence. The Computational Biology Research Group at the University of Washington has developed state of the art software to accomplish this. The difficulty is, there are thousands of distinct proteins found in rice. This presents a computational challenge that a single computer cannot solve within a reasonable timeframe. Therefore, volunteers of World Community Grid are invited to assist in this daunting task. Through collaboration with agricultural researchers and farmers, the hope is to eventually improve global rice yields and quality.

Significance
Hunger and malnutrition are the top risks to health worldwide. Nearly 30 percent of the world's population suffers from some form of malnutrition [1]. Every year, 10 million people die of hunger and hunger-related diseases. In fact, more people die from hunger and malnutrition annually than from AIDS, malaria, and tuberculosis combined [2].

Rice is the main food staple of more than half the world's population. 20 percent of the total food energy intake for every man, woman, and child in the world comes from rice. In Asia alone, more than 2 billion people get up to 70 percent of their daily dietary energy from rice and its by-products [3].

Improving strains of rice to yield larger, more resilient, and nutritionally-optimized harvests will positively impact the lives of billions of people.

Approach
Making better strains of rice has traditionally been accomplished through cross breeding of strains to produce hybrids with the best features. However, this is limited to crossing strains with easily observable traits.

Complex traits (such as high yield, disease resistance, or nutrient content) come from complex biochemical interactions of individual component proteins. Identifying such proteins and understanding their properties and interactions gives farmers the opportunity to affect these traits in a refined manner by choosing more subtle candidates for cross breeding. Predicting the structure of proteins can provide insight into the roles they play in the biochemistry of these traits.

Как работает рассчетное ядро
ВЮ содержит некоторые исходные параметры состояния белков. Каждое новое вычисление берет исходные параметры и генерирует новые значения в пределах некоторой нормы, и каждое такое состояние считается примерно 2 минуты на среднем по мощности компьютере. После каждого рассчета ВЮ сохраняется (чекпоинт) и проверяет, превышен ли интервал в 6 часов. Если да - ВЮ отправляется на сервер. Иначе - считается дальше. Таким образом мощные компьютеры могут посчитать примерно 500 состояний за ВЮ, а слабые компьютеры около 200. Соответственно, мощные компьютеры получают больше очков за то же самое время.



График проекта


Це повідомлення відредагував Rilian: May 20 2010, 20:26

[Rilian]

A paper was published in the journal Bioinformatics, which describes the use of Graphics Processing Units (GPU's) to accelerate computations comparing protein structures.

Paper Title:

“Accelerated protein structure comparison using TM-score-GPU”

Lay Person Abstract:

As part of the analysis of the computed results of the Nutritious Rice for the World project, the researchers need to be able to compare protein structures and efficiently compute a similarity score. A scoring method based on “Template Modeling”, known as TM-score, provides significantly better results than the “root-mean-square-deviation” method, but requires much more computer processing time. To solve this problem the researchers developed a version of the TM-score algorithm which makes use of Graphics Processing Units (GPU's) which are found in newer video hardware, used particularly on gaming computers to enhance the visual experience. Using GPU's they were able to run millions of protein comparisons about 70 times faster. The paper describes how they accomplished this and they offer the software freely to other scientists, who may be able to use it for their research.

Technical Abstract:

Motivation: Accurate comparisons of different protein structures play important roles in structural biology, structure prediction and functional annotation. The root-mean-square-deviation (RMSD) after optimal superposition is the predominant measure of similarity due to the ease and speed of computation. However, global RMSD is dependent on the length of the protein and can be dominated by divergent loops that can obscure local regions of similarity. A more sophisticated measure of structure similarity, Template Modeling ™-score, avoids these problems, and it is one of the measures used by the community-wide experiments of critical assessment of protein structure prediction to compare predicted models with experimental structures. TM-score calculations are, however, much slower than RMSD calculations. We have therefore implemented a very fast version of TM-score for Graphical Processing Units (TM-score-GPU), using a new and novel hybrid Kabsch/quaternion method for calculating the optimal superposition and RMSD that is designed for parallel applications. This acceleration in speed allows TM-score to be used efficiently in computationally intensive applications such as for clustering of protein models and genomewide comparisons of structure.

Results: TM-score-GPU was applied to six sets of models from Nutritious Rice for the World for a total of 3 million comparisons. TM-score-GPU is 68 times faster on an ATI 5870 GPU, on average, than the original CPU single-threaded implementation on an AMD Phenom II 810 quad-core processor. Availability and implementation: The complete source, including the GPU code and the hybrid RMSD subroutine, can be downloaded and used without restriction at http://software.compbio.washington.edu/mis...nloads/tmscore/ . The implementation is in C++/OpenCL.

Access to Paper:

To view the paper, please click here.

http://www.worldcommunitygrid.org/about_us...o?articleId=209

[Rilian]

Just to let people know.

The way that science works is that publications are months if not years behind the current work. For example, the methodology described in the paper has already been incorporated into a new method for choosing the best protein structure using TM-score as a similarity measure. This has already been applied to the rice protein structures and we have already used the best structures to predict the function of the rice genes.

But to write the results up as papers, we have to do the control calculations and benchmarking. Then after writing the manuscripts, it goes to internal reviews, external reviews, revisions, proofs etc before it gets published.

However, we will be putting up the results on our website soon - right after we do the benchmarks.

Hong