A question for computational/medicinal chemists working in drug discovery: In your day-to-day work, which computational tools/programs (small or large; commercial or open source) do you find most useful? For which application(s)? Naming just one or two of your favorites would be much appreciated. Many thanks in advance!
It's always an interesting exercise to answer a question like this -- especially if one uses a fair number of different tools. After a bit of thought, I think I can narrow the list down to four programs that I find myself launching at least once (and often many times) in the course of most projects:
1) PyMol (http://sourceforge.net/projects/pymol/): in its earlier incarnations, the menu options were primitive and the command syntax was clunky, but this program has gotten increasingly friendly and I now count it as a great pal and confidante. Whereas I once only used this if I wanted to make fancy graphics, it is now my go-to tool for even the most routine structural visualization, and has an array of plug-ins and wizards that achieve exceptional analytical functionality.
2) Vega-ZZ (http://nova.disfarm.unimi.it/cms/index.php?Software_projects:VEGA_ZZ): old-school modelers might see this as a modeling environment with a feel and functionality that blends features of SYBYL and Insight-II. I am increasingly relying on Vega-ZZ for molecule building, editing, structural refinement, and conformational analysis, and have recently started making use of its molecular database functionality.
3) Weka (http://www.cs.waikato.ac.nz/ml/weka/): other users could probably suggest other data mining platforms with a greater array of supported methods for feature selection, classification and clustering, but I find that Weka is a great first step for chemoinformaticists who have realized the perils of over-fit linear regression QSAR models and are looking for an easy way to explore SAR relationships and to identify robust higher dimensional relationships.
4) PaDEL Descriptor (http://www.yapcwsoft.com/dd/padeldescriptor/): whenever I need a diverse array of molecular descriptors, this is my first choice. I realize that Dragon has a larger number of descriptors to choose from, but when it comes to using molecular features, I strongly prefer open-source. Why? Because if I can't figure out from the originating paper what a descriptor is supposed to represent, there's no better recourse than to stare at the source code for a while.