On third-party interfaces to 3DNA

One clear sign of 3DNA's acceptance by the nucleic-acid structure-related scientific community is its integration into a wide range of GUI/web-interfaces (e.g., NDB, w3DNA, 3D-DART etc) and many other packages. As the author and maintainer of the 3DNA software package, I am (of course) very happy to see this happen: I can easily imagine that better accessibility would make 3DNA open to an even larger community, including non-experts in nucleic acid structures (e.g., for educational purpose). It should be emphasized, however, that point-and-click interfaces to 3DNA (and any other serious scientific software, for that matter), while convenient to occasional users and for routine jobs, have their limitations.

Overall, a point-and-click interface is only sensible to well-defined, routine tasks. In the context of 3DNA, the following list of tasks (not necessarily complete) is perfectly suitable:

• Generate one of the 55 fiber models, where everything (model number, sequence or number of repeats) can be unambiguously defined through a web-form
• Build an arbitrary DNA model with a user-specific parameter set
  
• Create a blocview image for a nucleic-acid containing structure specified in PDB format
• Analyze a regular (i.e., not deviating too much) double helix structure, be it in A-, B-, or Z-form
• Calculate a list of backbone torsion angles and other parameters (with the "-s" option of find_pair and then analyze)
• Find a list of all possible (RNA) base-pairs fulfilling a specific set of geometric criteria (with the "-p" option of find_pair).
  

Other than those (and carefully selected/qualified other functionality), users should be very careful in what they get from running a third-party tool outputting 3DNA-related parameters. I know of the so many subtleties in using 3DNA to solve a specific problem, and most of the time the command-line driven style is the most effective, as it allows for easy automation of try-and-error iterations. Software developers who integrate 3DNA into their packages should make the limitations clear to their users, and direct any 3DNA-specific problems to the 3DNA forum.

In today's informatics world, there are so many "easy-to-use" tools available, claiming to be able to solve all sorts of problems (well, that's understandable -- otherwise, how could one get published, especially in the big journals/magazines?). Any serious scientist, however, should know what he/she is doing: it is easy to get some (magic) numbers by clicking a button, but understand clearly what the numbers mean is yet another story. I cannot emphasize more the importance of knowing one's tools, including their limitations.