ICM Manual

[ Search prosite | Binding site analysis | Pdb sequence generation | Pdb merge | Search pdb headers ]

How to search all Prosite patterns in your sequence

Use macro searchSeqProsite. For example: 
read pdb "2dhf" 
  make sequence a_1.1        # sequence of a PDB structure  
  show sequence 
  find prosite 2dhf_a        # 2dhf_a is the sequence of the protein
See also find prosite, find pattern and read prosite.

How to identify binding pockets


There are three algorithms (A, B, and C) with ICM which can identify pockets:
optiontargetmacro
A closed pockets`icmCavityFinder
B almost closed pockets--{make map potential} , etc., see below
C pockets with good ligand-binding potential`icmPocketFinder
For the areas of space attracting ligands (option C), use two macros:
Example: 
read pdb "1a28" 
 delete a_!1,2 
 convert  
 delete a_2 
 icmPocketFinder a_ 3. yes no

In the following example we find an almost closed pocket which can not be identified with icmCavityFinder . 
read pdb "1fm6"  # read the 'a' chain of RXR 
 delete a_!1,9    # keep the RXR and its ligand only 
 make map potential a_1 Box( a_ 1. ) 1.  # grid size 1.5 A 
 make grob m_atoms exact 0.1 solid  
 split g_atoms 
 cool a_ 
 display g_atoms2 reverse
If you have problems with identifying pockets, change the grid size, the threshold level for make grop m_atoms , or try to convert object to the ICM type (the conversion will add hydrogens and make the object more dense).


How to generate a non-redundant list of PDB sequences

The following script is a skeleton of the provided script _mkUniqPdbSeqs which is somewhat more automated. 
l_commands=no 
 errorAction="none"        # if something goes wrong do not  
                           # interrupt the loop 
 s_pdbDir = "/data/pdb/"   # make sure you have correct path 
 pdbDirStyle = 4           #  
 read sarray s_pdbDir+"/derived_data/index/source.idx" 
                          # you need a list of all pdb-entries  
                          # (4 char. code per line will do) 
 source = Tolower(Trim(Field(source,1))) 
 n=Nof(source) 
 for i=5,n 
   read pdb sequence resolution source[i] 
        # append resolution to the chain name (like 9lyz_a19) 
 endfor 
 group sequence "*" uniqSeqs unique 0.1 delete
        # cutoff inter-sequence  
        # distance 0.1 (dissimilar by more than 10%) 
# 
# Other possibilities 
# 
#  group sequence uniqSeqs unique 5     # if two seqs differ by more 
#                                       # than 5 mutations  
#  group sequence uniqSeqs unique       # throw away only identical  
#                                       # sequences  
# 
 
 write sequence s_inxDir + "/pdb1.seq" 
                               # actual sequences for searches 
 write Name(uniqSeqs) "chainList"               
                               # list of protein chains if you need it 
 quit

How to merge several pdb files

The simplest way to merge two pdb files is to read them as separate objects and the use the move a_1. a_2. command. Example: 
read pdb "1crn" 
 read pdb "1d48" 
 move a_2. a_1.        # merges objects 
 write pdb a_1. "both" # saves both files in pdb format 
 write object a_1.     # saves merged object in compact binary form

Before or after merging, the objects can also be edited, translated to a new position, rename chains, change residue numbers etc. Example: 
read pdb "1d48" 
 delete a_w* 
 delete a_2       # delete the second chain 
 read pdb "1crn" 
 delete a_/33:99  # delete a C-term. part of crambin 
 move a_1. a_2.   # merge the remains 
 write object a_

If you want to re-engineer a polypeptide chain of a protein, using two pdb-files, e.g. to transplant one part of a protein to another and restore the bonding connectivity, you may use the modify command: 
read pdb "1crn"  # one pdb 
 read pdb "1cbn"  # similar protein 
 modify a_1./20:25 a_2./20:25   
   # grafts a loop from 2nd object to the 1st one 
 write pdb a_1. "combo"


How to search headers of the PDB entries

There is an PDB.tab file which contains one line header descriptions of all the entries. Now you have three ways of doing it: