-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hello YunLin-- > > I am using xplor-nih ( Xplor-nih version used 2.21, system RedHat > Linux 8.0 ) > for ensemble-based structure calculation of Intrinsic disordered > protein and I have several problems: > 1. Setting threads during parallel calculation > I had used the following command to set threads per node: > xplor -py ensemble.py -o 1.out -parallel -machines machinefile > -num_threads 2 > After generating the extended structure, there were always 4 > threads (the ensemble size is 4) threads per node, while we have only > two CPUs per node. How to set threads number so that there are 2 > threads per node? this depends on your machinefile: you need to make sure that each node appears only once in this file. Also, if you are using pbs you might try pbsxplor - a front-end helper distributed with recent Xplor-NIH versions. This command takes all the normal arguments, and also the -l argument of qsub. It writes a pbs script and calls qsub, and it understands the -num_threads, and determines how many Xplor-NIH instances should be run per node. An example invocation would be: pbsxplor -py -l nodes=50 -num_threads 2 -o script.out script.py > > 2. ShapePot and RAPPot > According to the paper [C.D. Schwieters et al. Progr. NMR Spectr. > 48, 47-62 (2006)], the function of ShapePot and RAPPot were to help > refinement calculations converge. Are they necessary or optional for > ensemble-based calculation? It depends. We found that they weren't really necessary when order parameter restraints were used, or when there are restraints on overall molecular shape (such as from saxs data). In the case of an disordered protein, the use of these terms would be hard to justify. best regards-- Charles -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Processed by Mailcrypt 3.5.8+ <http://mailcrypt.sourceforge.net/> iEYEARECAAYFAkrvPowACgkQPK2zrJwS/lZqZQCfQ8Vn+wzQ5oNCe4Kyg0P844m8 ui8AoIBVgjM4bqom2gDtLW+89yeKSPn3 =wpwG -----END PGP SIGNATURE-----
