Identifying good targets for making antibodies.

The idea is to identify regions that are highly antigenic and and specific to the protein under study. Looking at a multi species view that would mean that regions that are highly antigenic (immunogenic), conserved across orthologues but not conseved amond paralogs (homologues proteins in the same species). Also, the problem of cross reactivity could occur by identical stretches in non-related species. Thus, a blast is necessary on the entire genome to rule out the similar regions.

Finding antigenic regions
----------------------

Now in order that antibodies can react specifically to the regions in proteins, follwing properties are necessary.

1) The region should be solvent accessible
2) The region should be hydrophilic
3) The region should have a specific secondary/tertiary structure.

Such regions can be divided into two - 1) continuous stretches along the protein sequence and 2) Non-continuous stretches along the protein sequence but continuous in the tertiary structure (protein surface)

Other desired structural properties are - 1) presence of a proline (possibility of a cis confirmation), beta turn, and so forth.
3D structures would help identifying non-contiguous stretches.

Methods:
--------

1) Blast again Genome as well as PDB
2) SMART/Pfam
3) hydrophobicity/hydrophilicity index and
4) Secondary structure and slovent acessibility prediction (PHD)
5) Turn prediction and soforth.


Any suggestions are well come