"Niels Johan Christensen" <s991418@student.dtu.dk> wrote:
A good deal of my previous work (calculation of various electronic properties) is based on structures from non terminal patched MODELLER 7.7 models. To be consistent, I would like to introduce the terminal acetylation/amidation with MODELLER 7.7 and subsequently calculate properties for (fragments of) these models. (Regarding "consistency", I am mainly considering the fact that I get slightly different models from versions 7.7 and 8.1 despite using the same template/sequence. The only difference in the input conditions, is that in the former case I use TOP scipting while in the latter I use Python.)
Your reasoning is incorrect; the difference between Modeller 7v7 and 8v1 is that they were built with different Fortran compiler versions. Over time, the way the compiler vendors optimize code (e.g. the ordering of floating point operations) leads to very small changes in the results. You'll also notice that if you build models on a Windows machine, for example, that they'll differ from those built on a Linux box, for the same reason. The Python and TOP interfaces to Modeller give the same results (for equivalent scripts, on the same machine and Modeller version) and in fact are tested precisely to ensure this.
Your script doesn't work in Modeller 7v7 because the patch residue CT2 in ${LIB}/top_heav.lib incorrectly adds hydrogens. This was one of the bugs fixed in Modeller 8v1. You can fix it yourself by comparing the top_heav.lib file from the two versions, and correcting the 7v7 copy.
Ben Webb, Modeller Caretaker -- modeller-care@salilab.org http://www.salilab.org/modeller/ Modeller mail list: http://salilab.org/mailman/listinfo/modeller_usage