I'm looking for software to automatically design antibodies that target a given protein or antigen.

  • 1
    $\begingroup$ No, there does not. It is possible to do in silico tweaking of sequences to do some affinity maturation, for example, but I've never heard of a method for de novo CDR design. Nature is much better at doing that for us. Also, even if such software existed, the computing power required would be enormous - much much more than an individual computer. $\endgroup$
    – MattDMo
    Dec 20 '16 at 17:49

Much easier said than done!

If I had sufficient funds and time to pursue your goal what I would do is comb the Protein Data Bank for all the PDB files of interest. Surely you have some protein that interest you. Might consider doing the following:

  1. Figure out what family your protein belongs in, considering it's number and presence of motifs: coils, helices, beta sheets, etc...

  2. Do a literature review of any protein that has already been studied belonging to this family. So when a protein in this family catalyzes a reaction what are its flexible and restrained regions?

  3. Use a program like Autodock to dock your protein to a protein that its protein family might interact with.

  4. Calculate the interaction energies using Gromacs, NAMD, or Gaussian. You can do a zero time calculation of initial states or pursue a molecular dynamics approach as the proteins move randomly to their most stable configurations, or even a molecular mechanics approach. Which systems are most favorable?

  5. You got this far. Now what? But then none of these steps creates your candidates. It just gives you an idea of the physics involved in already structurally identified compounds. Yes, you can make point mutations in VMD and Pymol. But, if you want to add a new motif to an existing protein structure it will not work because more than 20 amino acids in a peptide do not fold properly.

To do some of the biological statistics, the research collaboratory of structural bioinformatics is right here: http://www.rcsb.org/pdb/home/home.do

You can parse through their folders with wget or curl in bash scripting at http://files.rcsb.org/

But then assuming you do figure out a candidate protein that doesn't exist yet. How are you going to make it in the lab? It appears that it would be better to do microbial studies. Maybe you can screen for a match with micro-array plates.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.