This version needs additional input that defines molecular fragements.
Atoms belonging to a molecular fragment have the same number tag after xyz coordinates.
Benefits of this approach is getting a set of input structures that have basic connectivity
intact, while sacrficing un-biased exploration of the potential energy surface.
In the first step, a geometric center (C) of each fragment is located, and relative coordinates
of each atom in fragment is determined. For each non-frozen fragment, a random atom (A) is chosen,
and a random vector (V) is generated. Length of said vector is equal to minimal distance value
added to randomly chosen fraction of distance parameter.
New point (P) for each fragment is set
by combining coordinates of atom A and vector V. In each fragment-point (P) fragment-center (C)
of randomly rotated fragment is placed. This procedure completely ignores initial relative
positions of each fragment.
This version is better suited for placement of large number of same fragemnts. Common example
includes placement of water molecules in solvation shells around the solute.
Articles published with this (similar) procedure in our group include:
"A computational study of the chlorination and hydroxylation of amines by hypochlorous acid"
D. Šakić, M. Hanževački, D. M. Smith, V. Vrček, Org. Biomol. Chem. 2015, 13, 11740 - 11752.
Help from Valerije Vrček (discussion), Hendrik Zipse (documentation, suggestions) and Vasily Korotenko (testing) is greatly appreciated. Please report all suggestion, changes, improvements, and usage to dsakic@pharma.hr