Theoretical modeling of the folding properties of proteins has been carried out at three different levels of resolution. 1) Distributions of amino acids in x-ray crystallographic structures from the Protein Data Bank (PDB) and general proteins from the Protein Investigation Resource (PIR) show that the large hydrophobic amino acids (PHE,TYR, TRP and MET) have very low occurrence levels of about 10%. The smaller hydrophobic and the hydrophilic amino acids have occurrences which approximate a random distribution. 2) Using the notion of a """"""""signal"""""""" for early protein folding caused by the occurrence of the large hydrophobic amino acids, a number of low resolution physical models were constructed in an attempt to understand the interaction between the distribution of these signals and the architecture of the folded protein. A preliminary logic program was developed using computer modeling to evaluate the ideas which had been expressed in the physical models. The big hydrophobics can be arranged spatially in many different ways. The choice of the spiral-like right-hand model of folding turns out to be the main determinant of the search combinatorics. The polypeptide path through 3-space as well as the structure of the amino acid sidechains can be represented in many different ways ranging from very discrete to continuously variable. 3) The discrete type of model with and without sidechains was represented on a three- dimensional lattice. Several different lattice spacings were investigated to see if any particular spacing had important static and dynamic properties. Rules for polypeptide motion in the lattice were investigated using conventional and logic programming techniques. The collaborators at ICOT are attempting to introduce parallelism of computation into their version of the logic programming model.
