I'm new to working with Omnia, but in a way it feels like coming home - only to the home I wish I'd had, not to the one I really knew.
Back when I was a postdoctoral researcher doing molecular simulations (~2005), a colleague would often, perhaps half-jokingly, say that 90% of our job was converting data from one file format into another. We would spend a huge amount of time poring over documentation trying to make sure we got format specifications just right in order to put together a one-off conversion script to get data into the right format to apply the tool we needed to take us on to the next step. And if we weren't manipulating file formats, we were poring over documentation to figure out how to migrate our simulation protocol from one simulation package into another to take advantage of the latest algorithm we wanted, which was only available in a simulation package we hadn't already been using. And even worse was if we wanted an analysis tool designed to cope with a third file format, requiring multiple file format conversions which could be nearly impossible to validate.
Unfortunately, I think a substantial portion of my time over the last 10 years has been devoted to solving those same problems again and again. The science, in some ways, has seemed to be the easy and fun part, while all of the file format and software usage issues provide perpetual tedious time-sinks.
With Omnia, I'm finally seeing the future arriving. With tools like ParmEd and MDTraj, we're now able to convert between a wide range of file formats, allowing easy access to more simulation packages, and more analysis options (which can now be implemented in a general, rather than package-specific, way). And if algorithms we need aren't available, OpenMM provides an easier and more flexible way to get up and running with them without having to switch to a totally different simulation package. And the community arising around Omnia means that it now makes more sense to justify writing sound, reusable code rather than one-off tools which solve our specific problem but will never be used again.
While my group's involvement with Omnia is still in its early days, it has already been useful in helping us automate setup of simulations of solutes in arbitrary solvent mixtures for ongoing studies in the lab, via our SolvationToolkit. And we're devoting substantial effort towards Omnia-related development, and shifting applications work towards using many of the Omnia tools.
I'm so excited to finally be moving away from reinventing the wheel and towards devoting more of my time to the actual science that excites me.