SCHEMA algorithm for protein structure-guided recombination written for Python 2.7 by Ian Campbell and Josh Atkinson in the Silberg Lab at Rice University
The all_disruptions.py takes in two data sets: (1) PDB structure file and (2) a FASTA-formatted alignment file
The alignment file must be in fasta format and not have white spaces between residues. It must have an alignment of the structure sequence and the two parent chimera sequences. The included sequence_grabber.py script can obtain the sequence from the pdb file. I often use MAFFT, selecting FASTA format, to make the alignments.
General inputs: python all_disruptions.py
Here is an example:
>AAAA TWYRAPSQIL >seq_one TWYRAPSQID >seq_three ARDVOARDVO
The crossover file can be made via cross_maker.py. Instructions
for use are included in comments within the cross_maker.py file.
You will have the option to give the length of residues for your
protein, the start and ending indices of the crossover, and name
for the file.
General inputs: (python) (function_this) (length) (crossover_start) (crossover_end) (name)
For example: "python cross_maker.py 10 2 8 T1" in the command line will yield
AABBBBBBCC
As a file named T1.
General notes: This script will generate disruption in the form of E. It can tolerate holes in sequence structure, as long as the pdb file follows the general convention of making its atom numbering sequentially continuous even when there are gaps in structure (I.e. the missing "atoms" are not counted in the file. If the final atom # is 1000, then there will be 1000 atom entries.)
It is generally best to take the sequence from the pdb file using the sequence_grabber.py script, rather than downloading the sequence file from the pdb website. Older pdb structures will have discrepancies between listed structure sequences and the sequences that are actually present in the structure file.
Older pdb files may have unforeseen complications and require more scrutiny.
How it works:
-
The pdb file is read. For each atom, the atom's number (index in chain), the atom type, the residue in which the atom resides (residue index), the parent residue's type, and the atom's coordinates are extracted. This information is contained in the list of lists known as coordinatess.
-
The align file is read. The residues from the the structure sequence and two parent sequences are sorted into a list of lists.
-
True beginning and true end of chain are extracted from align file. This is a small step that accounts for the holes at the beginning and end of structure sequences and helps set up indices for later steps.
-
Atom to res and res to atom dictionaries are made. Using the coordinates list of lists, two dictionaries are made. One that takes in atom number (position in chain) and gives out residue number (residue position in chain) and one that takes in residue number and gives out a list of the atom numbers it contains.
-
The distances between all atoms are calculated, using coordinates, and sorted into matrix "d".
-
Matrix d is used to make another matrix, contacts. Contacts is n x n where n is the number of atoms in the chain. All of contacts elements are 0 initially, and replaced with 1 if the two atoms which represent the indices of that element have a distance less than 4.5 A. Contacts therefore represents a binary accounting of all atom contacts less than 4.5 angstroms.
-
Matrix contacts is converted into res_contacts. What this means is that another n x n matrix is created. This time n represents the number of residues in structure chain. This matrix is also populated only with 0's. Again the res_contacts matrix is iterated over and for each element whose indices, which represent residues, have contacting atoms the 0 is replaced with a 1. This is done using dictionaries created in 3 and the contacts matrix
-
Residues that don't match between given sequence and sequences structures are accounted for. All of these residues' columns and rows are given a 0 in the res_contacts matrix.
-
Non-breakable interactions are accounted for. Non-breakable interactions are defined as those which are identical in both parents. This step is done using the align file list of lists. All of these residues' columns and rows are given a 0 in the res_contacts matrix.
-
The bottom half of the matrix is removed. I don't want to double count interactions, so in this step I remove the bottom left half of the matrix (below the diagonal) by giving all elements in that region 0.
-
Disruptions are counted. Using the crossover file, the regions in which the crossover occurs are counted in the residue contacts matrix and added up.