Open3DALIGN is an open-source software aimed at unsupervised molecular alignment. Open3DALIGN is a command-line tool which is operated by means of a few commands which can be entered interactively from a command prompt, read from a batch script or piped through standard input. If PyMOL is installed on the system while Open3DALIGN is being operated interactively, the setup of alignments can be followed in real time on PyMOL's viewport.
A typical workflow could be the following:
1. A dataset is imported as SDF file
2. Optionally, for each compound of the dataset a quenched molecular dynamics (QMD) conformational search is carried out to find the most stable conformers in a user-defined energy range from the global minimum
3. One or more template compounds are chosen, then the whole dataset is best-aligned to the template(s). The alignment is a performed rigidly, using for each compound of the dataset either the single conformation originally imported or the conformational databases previously obtained by QMD. Additionally, SDF conformational databases can be easily imported from external conformational engines such as MOE and OMEGA. In the latter case, the conformers which best fit the template(s) are chosen out of the conformational pool. Interestingly, conformational flexibility may also be taken into account for templates.
4. Finally, an SDF database constituted by the dataset compounds aligned to each of the templates is obtained
The QMD search is accomplished running a number of short molecular dynamics runs (using the MMFF94 force-field) followed by minimization (using the MMFF94s force-field); these computations are carried out by calling four programs belonging to the TINKER molecular mechanics suite (namely, analyze, dynamic, minimize and optimize) included in Open3DALIGN. MMFF94 atom types are automatically assigned to all compounds of the dataset with the help of SDF2XYZ2SDF, another tool of ours based on OpenBabel.
Alignments are computed in an atom-based fashion (by means of a novel algorithm inspired to the LAMDA algorithm by Richmond and co-workers), in a pharmacophore-based fashion using Pharao as the alignment engine, or finally in a mixed mode consisting in the combination of the latter two methods.
High computational performance is attained splitting the computational tasks over several threads, according to the number of CPUs available in the host running Open3DALIGN.
All computations can be restarted from the point from which they were eventually stopped, allowing most intensive alignment tasks to be completed also in the presence of time-limited computational resources.