Unittest structure similarity #14
Conversation
| @@ -252,34 +238,27 @@ def compute_irmsd_fast(self,izone=None,method='svd',cutoff=10,check=True): | |||
| # read the izone file | |||
| if izone is None: | |||
| resData = self.compute_izone(cutoff,save_file=False) | |||
| elif not os.path.isfile(izone): | |||
| resData = self.compute_izone(cutoff,save_file=True,filename=izone) | |||
| # elif not os.path.isfile(izone): | |||
There was a problem hiding this comment.
izone is used as both input file and output file, which is not unique. I left a TODO comment around that to add output file to the function.
| data_decoy_long, data_decoy_short = self.get_data_zone_backbone(self.decoy,resData,return_not_in_zone=True) | ||
| data_ref_long, data_ref_short = self.get_data_zone_backbone(self.ref,resData,return_not_in_zone=True) | ||
|
|
||
| atom_long = data_ref_long.intersection(data_decoy_long) |
There was a problem hiding this comment.
this intersection stuff does all the work ?
There was a problem hiding this comment.
It gets the common residues or atoms and ignore missing ones. It's actually checking if the sequences are consistent or not, but it cannot fix it if not consistent(even the former code cannot do this).
|
|
||
|
|
||
| def compute_izone(self,cutoff=5.0,save_file=True,filename=None): | ||
| def compute_izone(self,cutoff=10.0,save_file=True,filename=None): |
There was a problem hiding this comment.
It should be 10Å for iRMSD and 5Å for FNAT.
It's a bug but fine, the code invoking it resets the cutoff to 10.
| # create the sql | ||
| sql_decoy = pdb2sql(self.decoy,sqlfile='decoy.db') | ||
| sql_ref = pdb2sql(self.ref,sqlfile='ref.db') | ||
|
|
||
| # extract the pos of chains A | ||
| xyz_decoy_A = np.array(sql_decoy.get('x,y,z',chainID='A')) | ||
| xyz_ref_A = np.array(sql_ref.get('x,y,z',chainID='A')) | ||
| xyz_decoy_A = np.array(sql_decoy.get('x,y,z',chainID='A', name=backbone)) |
There was a problem hiding this comment.
Only backbone is needed in L-RMSD calculation, for both fitting and rmsd calculation. Here is the original reference https://doi.org/10.1002/prot.10393.
Note: This PR contains #11, #13.