Re: [PyMOL] CEalign parameter d0 and d1
Thank you Jason. Quyen On Wed, Apr 17, 2013 at 10:18 AM, Jason Vertrees < jason.vertr...@schrodinger.com> wrote: > Hi Quyen, > > It's been a few years since I wrote that but here is what I recall: > > D0 is a constant cutoff that represents the upper bound of dissimilarity > for the difference of two distance matrices A_i and B_j, rooted at > subsequences i to i+8 and j to j+8, respectively. Raising this value allows > for sloppier fragments being considered. > > D1 is a constant cutoff that represents the upper bound of dissimilarity > for the entire path being considered. In fact, it's equal to path_score / > (window_size * path_length). > > So, D0 ensure that incoming fragments are still good. D1 ensures that as > we extend the current alignment the total quality of the current alignment > remains good. > > After this, all N top scoring paths are stored in the path cache and the > best path is returned. > > Cheers, > > -- Jason > > > On Wed, Apr 17, 2013 at 3:34 AM, QT wrote: > >> Dear all, >> >> I have a question regarding CEalign d0 and d1 parameter. What exactly >> are their functions? >> >> I don't know enough C++ to decipher the algorithm but they are used >> findPath() in ccealignmodule.cpp. It seems that that d0 controls a >> similarity matrix and d1 the path length. A d1 below a certain number does >> indeed make cealign return RMSD over smaller set of residues and there is a >> lower bound for d1 where cealign will fail to align. D0 behaves >> similarly. The default is 2 and 3 for d0 and d1. I'm going to guess that >> those numbers are also dimensionless. >> >> Will it be useful to tune the parameter d0 and d1? >> >> Best, >> Quyen >> >> >> -- >> Precog is a next-generation analytics platform capable of advanced >> analytics on semi-structured data. The platform includes APIs for building >> apps and a phenomenal toolset for data science. Developers can use >> our toolset for easy data analysis & visualization. Get a free account! >> http://www2.precog.com/precogplatform/slashdotnewsletter >> ___ >> PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net) >> Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users >> Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net >> > > > > -- > Jason Vertrees, PhD > Director of Core Modeling Products > Schrödinger, Inc. > > (e) jason.vertr...@schrodinger.com > (o) +1 (603) 374-7120 > -- Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter___ PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net) Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net
Re: [PyMOL] CEalign parameter d0 and d1
Hi Quyen, It's been a few years since I wrote that but here is what I recall: D0 is a constant cutoff that represents the upper bound of dissimilarity for the difference of two distance matrices A_i and B_j, rooted at subsequences i to i+8 and j to j+8, respectively. Raising this value allows for sloppier fragments being considered. D1 is a constant cutoff that represents the upper bound of dissimilarity for the entire path being considered. In fact, it's equal to path_score / (window_size * path_length). So, D0 ensure that incoming fragments are still good. D1 ensures that as we extend the current alignment the total quality of the current alignment remains good. After this, all N top scoring paths are stored in the path cache and the best path is returned. Cheers, -- Jason On Wed, Apr 17, 2013 at 3:34 AM, QT wrote: > Dear all, > > I have a question regarding CEalign d0 and d1 parameter. What exactly are > their functions? > > I don't know enough C++ to decipher the algorithm but they are used > findPath() in ccealignmodule.cpp. It seems that that d0 controls a > similarity matrix and d1 the path length. A d1 below a certain number does > indeed make cealign return RMSD over smaller set of residues and there is a > lower bound for d1 where cealign will fail to align. D0 behaves > similarly. The default is 2 and 3 for d0 and d1. I'm going to guess that > those numbers are also dimensionless. > > Will it be useful to tune the parameter d0 and d1? > > Best, > Quyen > > > -- > Precog is a next-generation analytics platform capable of advanced > analytics on semi-structured data. The platform includes APIs for building > apps and a phenomenal toolset for data science. Developers can use > our toolset for easy data analysis & visualization. Get a free account! > http://www2.precog.com/precogplatform/slashdotnewsletter > ___ > PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net) > Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users > Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net > -- Jason Vertrees, PhD Director of Core Modeling Products Schrödinger, Inc. (e) jason.vertr...@schrodinger.com (o) +1 (603) 374-7120 -- Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter___ PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net) Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net
[PyMOL] CEalign parameter d0 and d1
Dear all, I have a question regarding CEalign d0 and d1 parameter. What exactly are their functions? I don't know enough C++ to decipher the algorithm but they are used findPath() in ccealignmodule.cpp. It seems that that d0 controls a similarity matrix and d1 the path length. A d1 below a certain number does indeed make cealign return RMSD over smaller set of residues and there is a lower bound for d1 where cealign will fail to align. D0 behaves similarly. The default is 2 and 3 for d0 and d1. I'm going to guess that those numbers are also dimensionless. Will it be useful to tune the parameter d0 and d1? Best, Quyen -- Precog is a next-generation analytics platform capable of advanced analytics on semi-structured data. The platform includes APIs for building apps and a phenomenal toolset for data science. Developers can use our toolset for easy data analysis & visualization. Get a free account! http://www2.precog.com/precogplatform/slashdotnewsletter___ PyMOL-users mailing list (PyMOL-users@lists.sourceforge.net) Info Page: https://lists.sourceforge.net/lists/listinfo/pymol-users Archives: http://www.mail-archive.com/pymol-users@lists.sourceforge.net
