Dear fellow PHENIX users,

My lab group has developed a new PHENIX-based tool for automated identification and building of solvent molecules in x-ray crystal structures.  The software, dubbed PeakProbe, and a basic README are available on GitHub:

https://github.com/paulsmith638/PeakProbe

Fair warning: the project is still undergoing development and the linked version generates excessive output and not all features are fully tested.  The feature extraction stage of the program is slow, figure about 1s per peak analyzed.

Currently, PeakProbe can do the following:

-- Analyze and remodel existing solvent

-- Build water/sulfate and a few ions de novo

-- Indicate where other species, such as glycerol or PEG, should be considered

-- Identify possible model errors such as missing alternate conformations

PeakProbe requires only an input structure PDB (with or without solvent) and a structure factor file in any format supported by PHENIX.  The program uses a model derived from analysis of over 2.2 million density points in the PDB in structures with limiting resolutions as low as 5.0A.  Features from these points are scaled/transformed to allow comparison of input peaks to well-defined feature distributions at any resolution followed by calculation of the relative likelihood of a given type of solvent existing at a given point.

Feel free to try PeakProbe and I welcome all feedback, bug reports, and questions.  Thanks again to Nigel, Nat, Tom and all of the PHENIX/CCTBX team for their help and for developing the battery of CCTBX tools upon which PeakProbe is based.

--Paul