In conclusion, A3D is an extremely versatile tool which aims to
approximate protein aggregation propensities in physiological-like
environments, where proteins are folded into globular structures,
able to oligomerize and can fluctuate between several conforma-
tions, thus reshaping the protein aggregation prediction scenario.
4 Notes
- Naccess server, accessible athttp://www.bioinf.manchester.ac.
uk/naccess/nac_intro.html, obtains a Surface Accessible Area
(SAA) as a spheric probe of 1.4 A ̊ radius, equivalent to a water
molecule that rolls over the Van der Waals surface of the
molecule. The contour traced by the probe center is the SAA
value, which is then normalized for an amino acid X consider-
ing the extended tripeptide Ala-X-Ala as SAAmax. A3D uses
normalized values of Naccess for its algorithm, named RSA
(relative surface area). - In the best case, the user will use a high quality 3D structure
obtained experimentally. Unfortunately, for some proteins or
even domains, this crucial information is not available. In these
situations, 3D modeling of protein structures represents a
suitable alternative approach to estimate the final structure.
A3D accept most of the models, if they fulfil the input
requirements. - The user should take note that, when running A3D upon a
solution NMR structure, the algorithm will consider the aver-
age structure. - If the targeted protein does not have either a well-defined
structure or a good quality 3D-model or it simply behaves as
an unfolded protein, the user should redirect the aggregation
propensity predictions to first generation tools, in which the
aggregation prone region is forecasted upon the primary
sequence. Examples of useful and powerful predictors are:
AGGRESCAN [34], Zyggregator [48], PASTA 2.0 [49], Fol-
dAmyloid [50], and TANGO [51]. - A significant number of X-ray solved structures in the PDB
display artificial multimeric structures due to polypeptide repe-
titions in the asymmetric unit of the protein crystal. In these
cases, the user should select only one of them by writing the
letter of the specific sequence after the PDB code. Example: for
the PDB XXXX sequence A, XXXX:A. - A3D will not take into account for the mathematical calcula-
tions residues or atoms missing on the 3D structure. As a
consequence, the exposure of certain amino acids could be
wrongly estimated by the algorithm. In such a way: (1) apolar
Predicting the Aggregation of Protein Structures 439