hindrance between amino acid side chains and suboptimal rotamer
configurations of residues in close vicinity [37]. Then, the server
starts to calculate the aggregation propensity of the repaired static
structure. A3D incorporates the possibility to forecast the impact of
structure fluctuations on the aggregation propensity of polypep-
tides by implementing the CABS-flex protocol [35]. CABS-flex is a
high-resolution modeling approach that generates a collection of
derived structures, covering the most representative backbone fluc-
tuations. The dynamic mode of A3D calculates the structurally
corrected aggregation propensity upon each conformer and only
gives as an output the most aggregation prone as a proxy of the
conformation that will likely drive the aggregation reaction in
solution.2.3 Input As an input, A3D only accepts 3D structures obtained from X-ray
diffraction, solution NMR, or modeling approaches in PDB format
(seeNotes 2– 4 ). A3D has different input requirements depending
on the run mode.
l For the standardstatic mode, which represents the vast majority
of A3D runs, files containing both monomeric or multimeric
protein structures are suitable; however, the A3D server is
capped to .pdb files built from a maximum of 20,000 atoms,
approximately 1000 amino acids (seeNote 5). The system
accepts entries with missing amino acids and sequence gaps (see
Note 6).
l A3Ddynamic mode is more restrictive in terms of input
requirements since multimeric and incomplete structures are
no longer accepted, due to incompatibilities with the implemen-
ted CABS-Flex algorithm version. Thus, structures must consist
of a single and continuous chain containing the complete set of
backbone atoms (N, Cα, C, and O). The sequence length is
restricted to a maximum of 400 amino acids.
3 Methods
3.1 Front Page
and Protein Input
A3D webserver has a clear and friendly display front page (Fig.2).
By using the top-side toolbar links, the user can check the proper-
ties of the implemented algorithm (“About”tab) and anticipate
troubles while running the server (“tutorial” tab). It is possible to
check the status of submitted predictions and to review previous
runs on-line (“queue” tab). On the right side, is the submission
box, where the user can load the targeted structure and select some
optional run settings.
A3D is linked to the Protein Data Bank (PDB) and entries can
be introduced by just writing the PDB code inside the “PDB code”
window. Alternatively, local .pdb files can be uploaded manually
using the “Browse” button.Predicting the Aggregation of Protein Structures 431