Guidelines for protein modelling
A simple checklist of items necessary to describe computer modelling
studies is presented. Each of the items is provided with a brief
description. Which items of the guidelines should be included in a
certain article depends on the used modelling strategy. Before applying
any computational analyses, one has to know the quality of used data as
computational predictions can be useless if the starting point is wrong
or severely biased.
Precalculated models for many proteins are available at ModBase (Pieper
et al., 2014) and SwissModel Repository (Bienert et al., 2017), however,
they do not provide full details as instructed in here. These models are
predicted with automated pipelines, thus the modelling procedure is not
optimized for each structure. Such models can still be useful for
various applications when experimental structures have not been
determined. When close enough structures are available, automatic models
may be of higher quality than an inexperienced modeler can achieve.
Users of any types of structural predictions must bear in mind and be
aware of limitations of such models.