<div dir="ltr">ah i see, yes i had no movers or optimizers ...Â <div><br></div><div>so now i'm trying to set up a RigidBodyMover but i'm struggling to create the rigid bodies. There seem to be two different ways of doing this, either IMP.atom.create_rigid_body or IMP.atom.setup_as_rigid_body.Â <div> <br></div><div>If I use create_rigid_body then when i later try to add restraints i get told "leaf 'A' ... does not have mass"</div><div><br></div><div>If I use setup_as_rigid_body i can add the restraints but i when i try to add it to the RigidBodyMover i get told "Rigid body passed to RigidBodyMover must be set to be optimized" even thoughÂ rb.set_coordinates_are_optimized(True)</div> <div><br></div><div>what is the function of choice here ?</div><div><br></div><div>thanks !</div><div><br></div><div>josh</div><div><br></div><div>------------------------------</div><div><br></div><div><div>def create_representation():</div> <div>Â Â m= IMP.Model()</div><div>Â Â all=IMP.atom.Hierarchy.setup_particle(IMP.Particle(m))</div><div>Â Â all.set_name("the universe")</div><div>Â Â # create a protein, represented as a set of connected balls of appropriate</div> <div>Â Â # radii and number, chose by the resolution parameter and the number of</div><div>Â Â # amino acids.</div><div>Â Â Â </div><div>Â Â def create_protein_from_pdbs(name, files):</div><div>Â Â Â Â Â </div><div>Â Â Â Â def create_from_pdb(file):</div> <div>Â Â Â Â Â Â sls=IMP.SetLogState(IMP.NONE)</div><div>Â Â Â Â Â Â datadir = os.getcwd()</div><div>Â Â Â Â Â Â print datadir</div><div><span class="" style="white-space:pre"> </span> Â Â t=IMP.atom.read_pdb( datadir+'/' + file, m,</div> <div>Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â IMP.atom.ATOMPDBSelector())</div><div>Â Â Â Â Â Â del sls</div><div>Â Â Â Â Â Â #IMP.atom.show_molecular_hierarchy(t)</div><div>Â Â Â Â Â Â c=IMP.atom.Chain(IMP.atom.get_by_type(t, IMP.atom.CHAIN_TYPE)[0])</div> <div>Â Â Â Â Â Â Â </div><div>Â Â Â Â Â Â # there is no reason to use all atoms, just approximate the pdb shape instead</div><div>Â Â Â Â Â Â s=IMP.atom.create_simplified_along_backbone(c,1)</div><div>Â Â Â Â Â Â Â </div><div> Â Â Â Â Â Â # make the simplified structure rigid</div><div>Â Â Â Â Â Â rb=IMP.atom.create_rigid_body(s)Â </div><div>Â Â Â Â Â Â #rb=IMP.atom.setup_as_rigid_body(s)Â </div><div>Â Â Â Â Â Â #rb=IMP.atom.create_rigid_body(c)</div> <div>Â Â Â Â Â Â rb.set_coordinates_are_optimized(True)</div><div>Â Â Â Â Â Â print rb.get_coordinates_are_optimized()</div><div>Â Â Â Â Â Â Â </div><div>Â Â Â Â Â Â return rb Â Â Â Â Â Â Â Â </div><div><br></div><div>Â Â Â Â h= create_from_pdb(files[0])</div> <div>Â Â Â Â h.set_name(name)</div><div>Â Â Â Â all.add_child(h)</div><div><br></div><div>Â Â create_protein_from_pdbs("A", [Firstpdb])</div></div><div><br></div><div>(m,all)= create_representation()<br></div></div> <div class="gmail_extra"><br><br><div class="gmail_quote">On 11 July 2014 19:42, <span dir="ltr"><<a href="mailto:imp-users-request@salilab.org" target="_blank">imp-users-request@salilab.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> Send IMP-users mailing list submissions to<br> Â Â Â Â <a href="mailto:imp-users@salilab.org">imp-users@salilab.org</a><br> <br> To subscribe or unsubscribe via the World Wide Web, visit<br> Â Â Â Â <a href="https://salilab.org/mailman/listinfo/imp-users" target="_blank">https://salilab.org/mailman/listinfo/imp-users</a><br> or, via email, send a message with subject or body 'help' to<br> Â Â Â Â <a href="mailto:imp-users-request@salilab.org">imp-users-request@salilab.org</a><br> <br> You can reach the person managing the list at<br> Â Â Â Â <a href="mailto:imp-users-owner@salilab.org">imp-users-owner@salilab.org</a><br> <br> When replying, please edit your Subject line so it is more specific<br> than "Re: Contents of IMP-users digest..."<br> <br> <br> Today's Topics:<br> <br> Â Â 1. getting the MCCGsampler to work (Josh Bullock)<br> Â Â 2. Re: getting the MCCGsampler to work (Ben Webb)<br> Â Â 3. Re: getting the MCCGsampler to work (Ben Webb)<br> <br> <br> ----------------------------------------------------------------------<br> <br> Message: 1<br> Date: Fri, 11 Jul 2014 13:12:01 +0100<br> From: Josh Bullock <<a href="mailto:jma.bullock@gmail.com">jma.bullock@gmail.com</a>><br> To: <a href="mailto:imp-users@salilab.org">imp-users@salilab.org</a><br> Subject: [IMP-users] getting the MCCGsampler to work<br> Message-ID:<br> Â Â Â Â <<a href="mailto:CAHh_40_B9Ciag3LMdDMgFLJf2MCx2-tGPp3tmMke9MQriHwMJA@mail.gmail.com">CAHh_40_B9Ciag3LMdDMgFLJf2MCx2-tGPp3tmMke9MQriHwMJA@mail.gmail.com</a>><br> Content-Type: text/plain; charset="utf-8"<br> <br> Hello again,<br> <br> So I have the same overall problem as before - creating an ensemble of a<br> 4-subunit complex using MSconnectivity restraints. Having visualised the<br> output (via RMF - thanks Barak :?) , it's clear that no matter how many<br> steps of CG or MC I put, the models do not change from their initial random<br> placement. I know that the restraints are present because the models are<br> evaluated and scored appropriately.<br> <br> So I saw on an old (2011) nup84 example that MCCG can't handle rigid<br> bodies, is this still the case ? If so, should I switch to the DOMINO<br> sampler ? or it does work and likely there's an error in my code ...<br> <br> Many thanks !<br> <br> Josh<br> <br> -------------------------------------------------<br> <br> import IMP<br> import IMP.atom<br> import IMP.rmf<br> import inspect<br> import IMP.container<br> import IMP.display<br> import IMP.statistics<br> #import IMP.example<br> import sys, math, os, optparse<br> import RMF<br> <br> from optparse import OptionParser<br> <br> <br> # Convert the arguments into strings and number<br> Firstpdb = str(sys.argv[1])<br> Secondpdb = str(sys.argv[2])<br> Thirdpdb = str(sys.argv[3])<br> Fourthpdb = str(sys.argv[4])<br> models = float(sys.argv[5])<br> <br> #*****************************************<br> <br> # the spring constant to use, it doesnt really matter<br> k=100<br> # the target resolution for the representation, this is used to specify how<br> detailed<br> # the representation used should be<br> resolution=300<br> # the box to perform everything<br> bb=IMP.algebra.BoundingBox3D(IMP.algebra.Vector3D(0,0,0),<br> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â IMP.algebra.Vector3D(100, 100, 100))<br> <br> <br> # this function creates the molecular hierarchies for the various involved<br> proteins<br> def create_representation():<br> Â Â m= IMP.Model()<br> Â Â all=IMP.atom.Hierarchy.setup_particle(IMP.Particle(m))<br> Â Â all.set_name("the universe")<br> Â Â # create a protein, represented as a set of connected balls of<br> appropriate<br> Â Â # radii and number, chose by the resolution parameter and the number of<br> Â Â # amino acids.<br> <br> Â Â def create_protein_from_pdbs(name, files):<br> <br> Â Â Â Â def create_from_pdb(file):<br> Â Â Â Â Â Â sls=IMP.SetLogState(IMP.NONE)<br> Â Â Â Â Â Â datadir = os.getcwd()<br> Â Â Â Â Â Â print datadir<br> Â Â t=IMP.atom.read_pdb( datadir+'/' + file, m,<br> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â IMP.atom.ATOMPDBSelector())<br> Â Â Â Â Â Â del sls<br> Â Â Â Â Â Â #IMP.atom.show_molecular_hierarchy(t)<br> Â Â Â Â Â Â c=IMP.atom.Chain(IMP.atom.get_by_type(t,<br> IMP.atom.CHAIN_TYPE)[0])<br> Â Â Â Â Â Â if c.get_number_of_children()==0:<br> Â Â Â Â Â Â Â Â IMP.atom.show_molecular_hierarchy(t)<br> Â Â Â Â Â Â # there is no reason to use all atoms, just approximate the pdb<br> shape instead<br> Â Â Â Â Â Â s=IMP.atom.create_simplified_along_backbone(c,<br> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â 1)<br> Â Â Â Â Â Â #IMP.atom.destroy(t)<br> Â Â Â Â Â Â # make the simplified structure rigid<br> Â Â Â Â Â Â rb=IMP.atom.create_rigid_body(s)<br> Â Â Â Â Â Â rb=IMP.atom.create_rigid_body(c)<br> Â Â Â Â Â Â rb.set_coordinates_are_optimized(True)<br> Â Â Â Â Â Â return s Â Â Â # <------- swapping c with s will give a coarse<br> grain representation - much faster !<br> # Â Â Â Â Â Â return c<br> <br> Â Â Â Â h= create_from_pdb(files[0])<br> Â Â Â Â h.set_name(name)<br> Â Â Â Â all.add_child(h)<br> <br> Â Â create_protein_from_pdbs("A", [Firstpdb])<br> Â Â create_protein_from_pdbs("B", [Secondpdb])<br> Â Â create_protein_from_pdbs("C", [Thirdpdb])<br> Â Â create_protein_from_pdbs("D", [Fourthpdb])<br> Â Â #create_protein_from_pdbs("C", ["rpt3_imp.pdb"])<br> Â Â return (m, all)<br> <br> # create the needed restraints and add them to the model<br> <br> def create_restraints(m, all):<br> Â Â def add_connectivity_restraint(s):<br> <br> Â Â Â Â tr= IMP.core.TableRefiner()<br> Â Â Â Â rps=[]<br> Â Â Â Â for sc in s:<br> Â Â Â Â Â Â ps= sc.get_selected_particles()<br> Â Â Â Â Â Â rps.append(ps[0])<br> Â Â Â Â Â Â tr.add_particle(ps[0], ps)<br> <br> Â Â Â Â # duplicate the IMP.atom.create_connectivity_restraint functionality<br> <br> Â Â Â Â score=<br> IMP.core.KClosePairsPairScore(IMP.core.HarmonicSphereDistancePairScore(0,1),tr)<br> <br> Â Â Â Â #ub = IMP.core.HarmonicUpperBound(1.0, 0.1)<br> Â Â Â Â #ss = IMP.core.DistancePairScore(ub)<br> <br> Â Â Â Â r= IMP.core.MSConnectivityRestraint(m,score)<br> <br> Â Â Â Â iA = r.add_type([rps[0]])<br> Â Â Â Â iB = r.add_type([rps[1]])<br> Â Â Â Â iC = r.add_type([rps[2]])<br> Â Â Â Â iD = r.add_type([rps[3]])<br> Â Â Â Â #n1 = r.add_composite([iA, iB])<br> Â Â Â Â n1 = r.add_composite([iA, iB, iC, iD])<br> Â Â Â Â n2 = r.add_composite([iA, iB],n1)<br> Â Â Â Â n3 = r.add_composite([iB, iD],n1)<br> Â Â Â Â n4 = r.add_composite([iA, iB, iC],n1)<br> Â Â Â Â n5 = r.add_composite([iB, iC, iD],n1)<br> <br> Â Â Â Â m.add_restraint(r)<br> <br> Â Â evr=IMP.atom.create_excluded_volume_restraint([all])<br> Â Â m.add_restraint(evr)<br> Â Â # a Selection allows for natural specification of what the restraints<br> act on<br> Â Â S= IMP.atom.Selection<br> Â Â sA=S(hierarchy=all, molecule="A")<br> Â Â sB=S(hierarchy=all, molecule="B")<br> Â Â sC=S(hierarchy=all, molecule="C")<br> Â Â sD=S(hierarchy=all, molecule="D")<br> Â Â add_connectivity_restraint([sA, sB, sC, sD])<br> <br> Â Â nbl = IMP.container.ClosePairContainer([all], 0, 2)<br> Â Â h = IMP.core.HarmonicLowerBound(0, 1)<br> Â Â sd = IMP.core.SphereDistancePairScore(h)<br> Â Â # use the lower bound on the inter-sphere distance to push the spheres<br> apart<br> Â Â nbr = IMP.container.PairsRestraint(sd, nbl)<br> Â Â m.add_restraint(nbr)<br> <br> <br> Â Â # r1 = IMP.core.ExcludedVolumeRestraint(all)<br> Â Â # m.add_restraint(r1)<br> <br> <br> # find acceptable conformations of the model<br> def get_conformations(m):<br> Â Â sampler= IMP.core.MCCGSampler(m)<br> Â Â sampler.set_bounding_box(bb)<br> Â Â # magic numbers, experiment with them and make them large enough for<br> things to work<br> Â Â sampler.set_number_of_conjugate_gradient_steps(200)<br> Â Â sampler.set_number_of_monte_carlo_steps(40)<br> Â Â sampler.set_number_of_attempts(models)<br> Â Â # We don't care to see the output from the sampler<br> Â Â #sampler.set_log_level(IMP.SILENT)<br> Â Â # return the IMP.ConfigurationSet storing all the found configurations<br> that<br> Â Â # meet the various restraint maximum scores.<br> Â Â cs= sampler.create_sample()<br> Â Â return cs<br> <br> <br> # cluster the conformations and write them to a file<br> def analyze_conformations(cs, all, gs):<br> Â Â # we want to cluster the configurations to make them easier to<br> understand<br> Â Â # in this case, the clustering is pretty meaningless<br> Â Â embed= IMP.statistics.ConfigurationSetXYZEmbedding(cs,<br> <br> Â IMP.container.ListSingletonContainer(IMP.atom.get_leaves(all)), True)<br> Â Â cluster= IMP.statistics.create_lloyds_kmeans(embed, 10, 10000)<br> Â Â # dump each cluster center to a file so it can be viewed.<br> Â Â for i in range(cluster.get_number_of_clusters()):<br> Â Â Â Â center= cluster.get_cluster_center(i)<br> Â Â Â Â cs.load_configuration(i)<br> Â Â Â Â h = IMP.atom.Hierarchy.get_children(all)<br> Â Â Â Â #tfn = IMP.create_temporary_file_name("cluster%d"%i, ".rmf")<br> Â Â Â Â huh = "./models/CLUSTER%d"%i<br> Â Â Â Â huh = huh +".rmf"<br> Â Â Â Â #print "file is", tfn<br> Â Â Â Â print "file is", huh<br> Â Â Â Â rh = RMF.create_rmf_file(huh)<br> <br> <br> Â Â Â Â IMP.rmf.add_hierarchies(rh, h)<br> <br> Â Â Â Â # add the current configuration to the file as frame 0<br> Â Â Â Â IMP.rmf.save_frame(rh)<br> <br> Â Â Â Â #for g in gs:<br> Â Â Â Â Â # Â rh.add_geometry(g)<br> <br> <br> #******************************************************************************************<br> # now do the actual work<br> <br> (m,all)= create_representation()<br> #IMP.atom.show_molecular_hierarchy(all)<br> create_restraints(m, all)<br> <br> # in order to display the results, we need something that maps the<br> particles onto<br> # geometric objets. The IMP.display.Geometry objects do this mapping.<br> # IMP.display.XYZRGeometry map an IMP.core.XYZR particle onto a sphere<br> gs=[]<br> for i in range(all.get_number_of_children()):<br> Â Â color= IMP.display.get_display_color(i)<br> Â Â n= all.get_child(i)<br> Â Â name= n.get_name()<br> Â Â g= IMP.atom.HierarchyGeometry(n)<br> Â Â g.set_color(color)<br> Â Â gs.append(g)<br> <br> cs= get_conformations(m)<br> <br> print "found", cs.get_number_of_configurations(), "solutions"<br> <br> ListScores = []<br> for i in range(0, cs.get_number_of_configurations()):<br> Â Â Â Â cs.load_configuration(i)<br> Â Â Â Â # print the configuration<br> Â Â Â Â print "solution number: ",i,"scored :", m.evaluate(False)<br> Â Â Â Â ListScores.append(m.evaluate(False))<br> <br> # for each of the configuration, dump it to a file to view in pymol<br> for i in range(0, cs.get_number_of_configurations()):<br> Â Â cs.load_configuration(i)<br> Â Â h = IMP.atom.Hierarchy.get_children(all)<br> Â Â #tfn = IMP.create_temporary_file_name("josh%d"%i, ".rmf")<br> Â Â #print "file is", tfn<br> Â Â huh = "./models/IMP%d"%i<br> Â Â huh = huh +".rmf"<br> Â Â print "file is", huh<br> Â Â rh = RMF.create_rmf_file(huh)<br> <br> Â Â # add the hierarchy to the file<br> Â Â IMP.rmf.add_hierarchies(rh, h)<br> <br> Â Â # add the current configuration to the file as frame 0<br> Â Â IMP.rmf.save_frame(rh)<br> <br> Â Â #for g in gs:<br> Â Â Â # Â w.add_geometry(g)<br> <br> analyze_conformations(cs, all, gs)<br> -------------- next part --------------<br> An HTML attachment was scrubbed...<br> URL: <<a href="http://salilab.org/archives/imp-users/attachments/20140711/fc50f2fe/attachment.html" target="_blank">http://salilab.org/archives/imp-users/attachments/20140711/fc50f2fe/attachment.html</a>><br> <br> ------------------------------<br> <br> Message: 2<br> Date: Fri, 11 Jul 2014 11:26:24 -0700<br> From: Ben Webb <<a href="mailto:ben@salilab.org">ben@salilab.org</a>><br> To: Help and discussion for users of IMP <<a href="mailto:imp-users@salilab.org">imp-users@salilab.org</a>><br> Subject: Re: [IMP-users] getting the MCCGsampler to work<br> Message-ID: <<a href="mailto:53C02C50.4020507@salilab.org">53C02C50.4020507@salilab.org</a>><br> Content-Type: text/plain; charset=ISO-8859-1; format=flowed<br> <br> On 7/11/14, 5:12 AM, Josh Bullock wrote:<br> > So I have the same overall problem as before - creating an ensemble of a<br> > 4-subunit complex using MSconnectivity restraints. Having visualised the<br> > output (via RMF - thanks Barak :?) , it's clear that no matter how many<br> > steps of CG or MC I put, the models do not change from their initial<br> > random placement. I know that the restraints are present because the<br> > models are evaluated and scored appropriately.<br> <br> A sampler that ignores the scoring function would be pretty useless!<br> Obviously MCCG doesn't do that. So my guess would be that you didn't add<br> the restraints to the scoring function, the thresholds are all wrong (so<br> that no MC moves are ever accepted), you have terrible starting<br> conditions (so that the optimizer can never improve the score) or you<br> have a poor choice of MC move set. I don't see you actually adding any<br> Movers to your sampler in the script, so I'd suspect the last one. The<br> underlying MonteCarlo class keeps some statistics (e.g. number of<br> accepted/rejected moves) which might shed some light on what's going on.<br> <br> > So I saw on an old (2011) nup84 example that MCCG can't handle rigid<br> > bodies, is this still the case ?<br> <br> No. You just need to add a suitable RigidBodyMover.<br> <br> > If so, should I switch to the DOMINO<br> > sampler ?<br> <br> DOMINO enumerates in a discrete space, so is a rather different kind of<br> optimizer. It is probably not the best choice in your case.<br> <br> Â Â Â Â Ben<br> --<br> <a href="mailto:ben@salilab.org">ben@salilab.org</a> Â Â Â Â Â Â Â Â Â Â Â <a href="http://salilab.org/~ben/" target="_blank">http://salilab.org/~ben/</a><br> "It is a capital mistake to theorize before one has data."<br> Â Â Â Â - Sir Arthur Conan Doyle<br> <br> <br> ------------------------------<br> <br> Message: 3<br> Date: Fri, 11 Jul 2014 11:42:19 -0700<br> From: Ben Webb <<a href="mailto:ben@salilab.org">ben@salilab.org</a>><br> To: Help and discussion for users of IMP <<a href="mailto:imp-users@salilab.org">imp-users@salilab.org</a>><br> Subject: Re: [IMP-users] getting the MCCGsampler to work<br> Message-ID: <<a href="mailto:53C0300B.5010504@salilab.org">53C0300B.5010504@salilab.org</a>><br> Content-Type: text/plain; charset=ISO-8859-1; format=flowed<br> <br> To clarify a little, the MCCGSampler tries to wrap a common use of the<br> underlying MonteCarlo optimizer (that use case being a bunch of<br> independently movable x,y,z particles). If in your case you actually<br> have a bunch of rigid bodies, you should probably use the MonteCarlo<br> optimizer directly and add RigidBodyMovers to it.<br> <br> Â Â Â Â Ben<br> --<br> <a href="mailto:ben@salilab.org">ben@salilab.org</a> Â Â Â Â Â Â Â Â Â Â Â <a href="http://salilab.org/~ben/" target="_blank">http://salilab.org/~ben/</a><br> "It is a capital mistake to theorize before one has data."<br> Â Â Â Â - Sir Arthur Conan Doyle<br> <br> <br> ------------------------------<br> <br> _______________________________________________<br> IMP-users mailing list<br> <a href="mailto:IMP-users@salilab.org">IMP-users@salilab.org</a><br> <a href="https://salilab.org/mailman/listinfo/imp-users" target="_blank">https://salilab.org/mailman/listinfo/imp-users</a><br> <br> <br> End of IMP-users Digest, Vol 38, Issue 15<br> *****************************************<br> </blockquote></div><br></div></div>