PCA-app FIT contact map core option

[Barry Grant]

Will give a non sensible core region when analyzing SERCA family structures. E.g. Try with PDB ID 1SU4 and compare fir with new and old core approaches.

[Xinqiu Yao]

Sorry for late response.

The problem was caused by the multi-domain architecture of SERCA and the fact that one of the cytoplasmic domain is more stable than the TM domain. The core.find2() function by default reports the contact core "cluster" with the highest intra-domain stability, but it may generate high overall structural variance if these cores are located in a small globular domain (such as the case of SERCA). Fortunately, the algorithm implemented in the function inherently finds other possible (domain based) cores. One solution is: We inspect multiple core candidates (in the case of multi-domain proteins), do one-time structure fitting and "ellipsoid error volume" calculation, and output the cores associated with the smallest total error volume. (Please see the latest commit here for these changes). For single domain proteins, the function is the same as before.

Let me know if you like these updates or not!

[Xinqiu Yao]

Solved by recent commit (Check by inputting '1SU4' on the web server).