A combined threading and genetic algorithm based approach to predict protein structure

We present new approaches and programs in our study of protein tertiary structure prediction based on the genetic algorithm (GA), applying also knowledge guided and database guided methods. Common desktop computers are able to run our new software which makes its application possible in the laboratory and in student teaching. <br> <br>We introduce a new fitness function weight estimate for the genetic algorithm based on similarity of predicted secondary element content to known x-ray crystal structures from the Brookhaven Protein Database. In analogy to existing strategies we call this principle "threading-GA". This allows us to assist previous folding routines with a more natural approach. The evaluation routines are included within the application software so that future crystal structures may be easily included later and directly by the user. <br> <br>As examples we show the simulation for several proteins. Some of them, like small toxin proteins, are of medical importance and should underline the need of prediction software in this research field. In times where protein sequence information is growing daily, tools are needed to derive structure predictions from it. Our software may be directly used for ab initio prediction of protein structure from sequence. In addition, the user may add specific information on the protein in question. This increases the quality of the resulting model. <br> <br>We show further that a GA may have the potential to serve in prediction of structures for helical transmembrane proteins. Here we combine a systematic plot of the Ramachandran protein conformation map with our existing application. This setting allows us to generate models with correct membrane topology. However, further investigation will be necessary on that challenging topic. <br> <br>The core GA has been optimized for the specific task of protein structure prediction. The program is now able to simulate small protein structures in less than a minute. Our software is completely written in ANSI C programming language. A system independent interface allows our application to run on the major operating systems.

Author dc:creator

• Hench, Jürgen Christian Hans

Contributors dc:contributor

• Dandekar, Thomas

• Proteinstruktur
• Vorhersage
• GA
• genetischer
• Algorithmus
• protein
• structure
• prediction
• genetic
• algorithm

Repository record source_url

https://freidok.uni-freiburg.de/data/2045

OAI identifier oai:identifier

oai:freidok.uni-freiburg.de:2045