利用粒子群算法在菱形网格上预测蛋白质结构

本文在菱形网格上研究讨论了二维HP模型。首先,将蛋白质结构预测问题转化成一个数学问题,并简化成氨基酸序列中每个氨基酸与网格格点的匹配问题。为了解决这个数学问题,我们改进并扩展了经典的粒子群算法。为了验证算法和模型的有效性,我们对一些典型的算例进行数值模拟。通过与方格网上得到的蛋白质构象进行比较,菱形网上的蛋白质构象更自然,更接近真实。我们进一步比较了菱形网格上的紧致构象和非紧致构象。结果显示我们的模型和算法在菱形网格上预测氨基酸序列的蛋白质结构是有效的有意义的。

Predicting protein structure on rhombus lattice by particle swarm optimization

Successfully predicting protein structures is very significant in exploring the activity of life,and protein structure is decided by amino acid sequence. In order to solve the core problem of field of biology, we studied the two-dimensional hydrophobic-hydrophilic model on rhombus lattice. Firstly, protein structure prediction problem was transformed as a mathematical optimization problem and abstracted as a match problem between amino acids and lattice vertexes. To solve this problem, classical Particle Swarm Optimization algorithm was extended and improved. Then several benchmark examples were simulated. Compared with configurations on square lattice, conformations on rhombus lattice were satisfied with more biological character. We further compared the compact conformations with non-compact conformations on rhombus lattice. The results indicated that our model and method were effective and significant on rhombus lattice.