Biology Reference
In-Depth Information
4. Protein Model Quality
4.1. Correctness and Accuracy
Conceptually, the quality of a model can be determined by two distinct
criteria that will determine its applicability to solve biological questions:
the correctness of a model, which is largely dictated by the quality of the
sequence alignment used to guide the modeling process; and the accuracy
of a model, which is essentially limited by the deviation of the used
template structure(s) from the target structure.
4.1.1. Model correctness
The comparative protein structure modeling methods implemented in
SWISS-MODEL rely on the evolutionary relationship between target
and template proteins. As sequence similarities reflect these evolutionary
relationships, they can be used as a first approximation of structural rela-
tedness. Consequently, the two major limitations of sequence comparison
methods are carried over to comparative protein modeling. Firstly, there
might simply not be any sequence of known 3D structure with enough
similarity to the target sequence; and secondly, sequence comparison
algorithms are known to yield erroneous alignments as the level of iden-
tity between sequences decreases below 30%. The former limitation sim-
ply precludes the building of a comparative model, while the latter is the
cause of incorrect models which must be considered with caution or in
many cases simply discarded. Indeed, if the sequence alignment is wrong
in at least one region of the protein, then the spatial arrangement of the
residues in this portion of the model will be incorrect and furthermore
adversely impact the conformation of neighboring residues.
Sequence comparison algorithms have become so sensitive that today
they can detect very distantly related sequences and allow fold assign-
ments of distantly related proteins. 25 From the perspective of a structural
biologist, however, amino acid sequences are little more than a string-
based notation attempting to represent a complex structure, analogous to
SMILES which represents small molecular entities as strings of characters.
By extension, sequence alignments are highly simplified representations
Search WWH ::




Custom Search