June 2020 - modeller_usage

Multiple CPUs
by Casey Becker 06 Jul '20

06 Jul '20

Hello again, I'm trying to use modeller directly through python's IDLE interface to run a parallel job. I've attached my script and also the python core file that's created up to a point, because it stops before it actually makes the models. Three files with .slave0, .slave1, and .slave2 are created, but when edited with notepad, show an error. I've attached one of them as well. Thanks.

2 1

Treading onto known PDB template
by Geoffrey Woollard 06 Jul '20

06 Jul '20

I'm studying a transmembrane ion channel (protein 1). There are structures of protein 1 in a closed conformation, and I am interested in the open conformation. There is a very similar protein in an open conformation (protein 2). So I am trying to thread the sequence of protein 1 onto the atomic model of protein 2. I know the template. I tried robetta and swiss prot for this. How should I do things with modeller? Could someone point me to a tutorial? The sequences are quite similar, except for some insertions. I know my protein very well and am looking for a protein modelling approach where I can have low level control, and manual manipulation at the sequence alignment and structure level. I don't need something that does everything automagically. -- Geoffrey Woollard, MSc PhD student Department of Medical Biophysics, University of Toronto Princess Margaret Cancer Centre, University Health Network

2 2

Re: [modeller_usage] z-DOPE score
by Mariana GONZALEZ MEDINA 22 Jun '20

22 Jun '20

Makes sense now, thank you! Mariana ________________________________ From: Min-yi Shen <minyishen(a)gmail.com> Sent: Friday, June 19, 2020 5:27:08 PM To: Modeller Caretaker Cc: Mariana GONZALEZ MEDINA; modeller_usage(a)salilab.org Subject: Re: [modeller_usage] z-DOPE score Original DOPE author here. The Z-dope is effectively an estimated z-score of your protein model based on the protein composition and size, as the raw DOPE scores are size and protein dependent. Z-dope used a linear model to estimate the mean and sd of a protein’s DOPE score based on their sizes and compositions. It was trained by millions of models from ModBase. -1.5 means your model’s DOPE score is -1.5 standard deviation lower than the average... usually a good indicator of good native overlap. I can’t believe I can still recall all these... On Thu, Jun 18, 2020 at 11:59 Modeller Caretaker <modeller-care(a)salilab.org<mailto:modeller-care@salilab.org>> wrote: On 6/18/20 9:48 AM, Mariana GONZALEZ MEDINA wrote: > I have a question regarding the zDOPE score. I have read some paper that > state this: "The Z-DOPE scores ranged from −1.63 to −1.85, where a score > of less than −1 indicates a “reliable” model (i.e., 80% of its Cα atoms > are within 3.5 Å of their correct positions)" > > I am wondering how is it that a score of less than -1 indicates that 80% > of its Cα atoms are within 3.5 Å of their correct positions? I would say "implies" rather than "indicates". DOPE is a model evaluation function. Of course we don't know the true structure so we can only guess with some degree of confidence (the more negative the score, the more confident I would be) whether a good DOPE score really means a correct model. Benchmarking does show that this is typically the case though; see the DOPE paper. Ben Webb, Modeller Caretaker -- modeller-care(a)salilab.org<mailto:modeller-care@salilab.org> https://salilab.org/modeller/ Modeller mail list: https://salilab.org/mailman/listinfo/modeller_usage _______________________________________________ modeller_usage mailing list modeller_usage(a)salilab.org<mailto:modeller_usage@salilab.org> https://salilab.org/mailman/listinfo/modeller_usage

1 0

z-DOPE score
by Mariana GONZALEZ MEDINA 18 Jun '20

18 Jun '20

Hi, I have a question regarding the zDOPE score. I have read some paper that state this: "The Z-DOPE scores ranged from −1.63 to −1.85, where a score of less than −1 indicates a “reliable” model (i.e., 80% of its Cα atoms are within 3.5 Å of their correct positions)" I am wondering how is it that a score of less than -1 indicates that 80% of its Cα atoms are within 3.5 Å of their correct positions? How is it possible to know the percent of atoms in the "correct position" based on how small (or not) the z-DOPE is? What can I say about the magnitude of the negative values besides stating that since the models have negative values they are acceptable? Best, Mariana

2 1

Modeller Outputs TER field wrong in PDB
by Pedro Guillem 18 Jun '20

18 Jun '20

I had some issues today using some PDBs generated by Modeller, and I tracked the error to the fact the TER fields inside the PDB have more columns (which should not be there) For example ATOM 1393 C2' DT D 35 81.341 17.516 55.945 1.00 48.63 C ATOM 1394 C3' DT D 35 82.152 18.800 55.754 1.00 48.81 C ATOM 1395 O3' DT D 35 83.036 19.108 56.819 1.00 49.40 O *TER 1396 DT D 35* ATOM 1397 N SER A 20 33.719 48.929 107.140 1.00 49.89 N ATOM 1398 CA SER A 20 34.080 48.490 108.522 1.00 50.05 C ATOM 1399 CB SER A 20 33.010 48.924 109.527 1.00 50.13 C I understand that the TER separator is not supposed to have anything afterwards. Is there a particular reason why Modeller is writing my TER records like that? It should read only "TER" Best Pedro

2 3

Can I calculate the "Objective Function" based on a PDB without doing the comparative modelling?
by ZHANG Cheng 14 Jun '20

14 Jun '20

The "Objective Function" is automatically shown on the second line of remark after doing the MODELLER comparative modelling. Now I have a structure generated by RosettaCM, and I want to use the MODELLER's "Objective Function" to evaluate its reliability. Would it be possible? Thank you!

2 1

Can't really edit my script
by Casey Becker 11 Jun '20

11 Jun '20

Hello, I'm new to modeller and I'm eventually trying to make several hundred models using multiple cpus, however it seems I can't change anything in my script except number of models without causing the system to break and not work. No error pops up, it simply says the task completed instantaneously, doesn't make any models, and doesn't pop up the ok_model window. My script is below from modeller import * from modeller.automodel import * # ---------------------- namelist.dat -------------------------------- # A "namelist.dat" file contains the file names, which was output from # the Modeller dialog in Chimera based on user's selection. # The first line it the name of the target sequence, the remaining # lines are name of the template structures namelist = open( 'namelist.dat', 'r' ).read().split('\n') tarSeq = namelist[0] template = tuple( [ x.strip() for x in namelist[1:] if x != '' ] ) # ---------------------- namelist.dat -------------------------------- # This instructs Modeller to display all log output. log.verbose() # create a new Modeller environment env = environ() # Directory of atom/PDB/structure files. It is a relative path, inside # a temp folder generated by Chimera. User can modify it and add their # absolute path containing the structure files. env.io.atom_files_directory = ['.', './template_struc'] # create a subclass of automodel or loopmodel, MyModel. # user can further modify the MyModel class, to override certain routine. class MyModel(automodel): def customised_function(self): pass #code overrides the special_restraints method #def special_restraints(self, aln): #code overrides the special_patches method. # e.g. to include the addtional disulfides. #def special_patches(self, aln): a = MyModel(env, sequence = tarSeq, # alignment file with template codes and target sequence alnfile = 'alignment.ali', # PDB codes of the templates knowns = template) # index of the first model a.starting_model = 1 # index of the last model a.ending_model = 5 loopRefinement = False # Assesses the quality of models using # the DOPE (Discrete Optimized Protein Energy) method (Shen & Sali 2006) # and the GA341 method (Melo et al 2002, John & Sali 2003) a.assess_methods = (assess.GA341, assess.normalized_dope) # ------------------------- build all models -------------------------- a.make() I've tried changing the MyModel part to just be automodel, because I was told it was a custom subclass and needed special treatment, which broke the code. I also tried changing the knowns = template to knowns = my template ID. It also broke the code. Thanks.

2 1

Parallel job
by Casey Becker 09 Jun '20

09 Jun '20

Hello, I'm fairly new to modeller, and I'm trying to run a parallel job. I can't seem to get the 'from modeller.parallel import *' line to work. When I input it, no models are created. I have python 3.8 installed, and I'm not sure why it's happening. My script is below. from modeller import * from modeller.automodel import * from modeller.parallel import * j= job() j.append(local_slave()) j.append(local_slave()) j.append(local_slave()) log.verbose() env = environ() env.io.atom_files_directory = ['.', './template_struc'] # create a subclass of automodel or loopmodel, MyModel. # user can further modify the MyModel class, to override certain routine. class MyModel(automodel): def customised_function(self): pass #code overrides the special_restraints method #def special_restraints(self, aln): #code overrides the special_patches method. # e.g. to include the addtional disulfides. #def special_patches(self, aln): a = MyModel(env, sequence = tarSeq, # alignment file with template codes and target sequence alnfile = 'alignment.ali', # PDB codes of the templates knowns = template) # index of the first model a.starting_model = 1 # index of the last model a.ending_model = 100 loopRefinement = False # Assesses the quality of models using # the DOPE (Discrete Optimized Protein Energy) method (Shen & Sali 2006) # and the GA341 method (Melo et al 2002, John & Sali 2003) a.assess_methods = (assess.GA341, assess.normalized_dope) a.use_parallel_job(j) a.make() I'm still figuring my way out around this. Any help appreciated. Thanks.

2 1

Identical output for different batches
by ZHANG Cheng 08 Jun '20

08 Jun '20

Dear MODELLER, In order to generate hundreds of PDBs, I submit multiple jobs to the cluster. The input template is the same for all the jobs. For each job (i.e. each batch), 10 PDBs are generated. The 10 PDBs within the same batch are different in their "Objective Function" in the PDB remark. But the PDBs between the batches have the same "Objective Function". For example,  ) the 3rd PDB in Batch 1 has the same "Objective Function" as the 3rd PDB in Batch 5 ) the 6th PDB in Batch 2 has the same "Objective Function" as the 6th PDB in Batch 10 So it seems that all the jobs are started from a same "seed". Is there a way to randomise the seed? Thank you! Cheng The script I use is: from modeller import * from modeller.automodel import *    # Load the automodel class log.verbose() class MyModel(automodel):     def special_patches(self, aln):         # Rename both chains and renumber the residues in each         self.rename_segments(segment_ids=['A', 'B', 'C', 'D','E', 'F','G', 'H','I', 'J','K', 'L','N', 'O','M', 'P','Q', 'R','S', 'U','V', '1','2', '3','4', '5','6', '7','8', '9',],                              renumber_residues=[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1])         # Another way to label individual chains:         self.chains[0].name = 'A'         self.chains[1].name = 'B'         self.chains[2].name = 'C'         self.chains[3].name = 'D'         self.chains[4].name = 'E'         self.chains[5].name = 'F'         self.chains[6].name = 'G'         self.chains[7].name = 'H'         self.chains[8].name = 'I'         self.chains[9].name = 'J'         self.chains[10].name = 'K'         self.chains[11].name = 'L'         self.chains[12].name = 'N'         self.chains[13].name = 'O'         self.chains[14].name = 'M'         self.chains[15].name = 'P'         self.chains[16].name = 'Q'         self.chains[17].name = 'R'         self.chains[18].name = 'S'         self.chains[19].name = 'U'         self.chains[20].name = 'V'         self.chains[21].name = '1'         self.chains[22].name = '2'         self.chains[23].name = '3'         self.chains[24].name = '4'         self.chains[25].name = '5'         self.chains[26].name = '6'         self.chains[27].name = '7'         self.chains[28].name = '8'         self.chains[29].name = '9' env = environ() # directories for input atom files env.io.atom_files_directory = ['.', '../atom_files'] # Be sure to use 'MyModel' rather than 'automodel' here! a = MyModel(env,             alnfile  = '../../1_raw/JN254802-5tx1.ali' ,     # alignment filename             knowns   = '5tx1',              # codes of the templates             sequence = 'JN254802')              # code of the target a.starting_model= 1                # index of the first model a.ending_model  = 10                # index of the last model                                    # (determines how many models to calculate) a.make()                           # do comparative modeling

2 3