7 Combinatories and Topology of theβ-Sandwich andβ-Barrel Proteins 129The search is carried out with predefined sets of several (8–12) sequence
determinants, instead of analysis of a whole protein sequences.
We focus in this work on the analysis of two large groups of theβ-proteins
sandwich-like and the barrel-like proteins. The goal of this research is to define
the structural and sequence features, which these very different proteins have
in common. Our first task is to analyze the supersecondary substructure of
proteins to determine whether they have features that are invariant. Another
aspect of our research involves finding conserved positions in sequences that
are occupied by similar residues in all proteins.
Analysis of the supersecondary structure is based on the information about
hydrogen bonds contacts between the main chain atoms of residues. It presents
as a list of number of residues, which are connected by the hydrogen bonds
between the main chain atoms of residues. It turned out that examination
of only the list of the numbers of the hydrogen bonded residues was suffi-
cient to determine: (a) a secondary structure; (b) an arrangement of strands
(a supersecondary structure); (c) a protein fold (structural classification of the
protein); (d) a set of the rules that governs the arrangement of the strands
in the barrel and sandwich structures; (e) the supersecondary patterns: two
pairs of strands, whose location in the structure is common in all barrel and
sandwich proteins. This analysis discovered that despite a seemingly unlimited
number of arrangements of strands there exists a rigorously defined constraint
on supersecondary structures.
Another aspect of our research involves finding positions in sequences that
are occupied by similar residues. The problem of the discovery of conserved
positions is not easy for highly various groups of proteins as sandwich and
barrel proteins. A comparison of the amino acid sequences in various super-
families showed that the sequences are so diverse that even the most powerful
approaches such as PSI-BLAST and hidden Markov model cannot find any
sequence homology. However, the delineation of the invariant supersecondary,
substructure, common for different superfamilies and protein folds, makes pos-
sible a secondary structure-based multialignment. It results in the set of the
key conserved positions, whose residues share both structural and chemical
properties and have a decisive role in geometry of the beta proteins.
Thus putting the information about amino acid sequences and hydrogen
bond contacts together we give in account the principal relations between
sequence and structure in the beta proteins.
7.2 Overview of the Structures
Allβ-proteins can be divided into several groups in relation to their “struc-
tural design” likeβ-sandwiches,β-barrels,β-propellers, and others [13]. Each
of these architectural designs encompasses a large number of protein fami-
lies, superfamilies, and folds within the structural hierarchical classification