I am new to programming and I am given the task to extract structural data of soluble proteins from PDB and then shortlist the soluble enzymes from the extracted data set for further classification based shortlisting of soluble non membrane enzymes. Since I do not have much expertise in programming I am finding it difficult to figure out how to solve the problem. Any help would be really appreciated.
$\begingroup$ It sounds very much that you are interested in enzyme properties that aren't the cartesian coordinates of the atoms. Sure, you can calculate the Gibbs free energy of folding of a structure, a proxy for stability, but converting this to melting temperature without a reference is crude and at that point you may as well use the ProTherm database. One can also do analyses of the Poisson-Boltzmann surface electrostatic charges etc. but for solubility simply calculating the isoelectric point from a primary sequence will be easier and will circumvent the need for a structure... $\endgroup$– Matteo FerlaDec 13, 2021 at 9:47
1$\begingroup$ @MatteoFerla Thank you for your response. Is there a way by which I can import the pdb files of proteins using tags like Soluble, Membrane/Non-Membrane ? The initial Import of soluble proteins from the database is an issue that i am facing. Once i am able to import all the proteins, then I plan to use Binary SVM to further classify then into enzymes and non enzymes $\endgroup$– Arinjoy DattaDec 13, 2021 at 17:34
$\begingroup$ The classification of enzyme vs. non-enzyme should definitely not be done with a classifier on extracted structural properties. Parsing the UniProt entry for the protein has all you need. Take an enzyme: uniprot.org/uniprot/P06721 there's the field in the XML (sorry no JSON) that gives you the EC number and another the subcellular locationation. It also gives what PFam domains are there. Alternatively one extract the GO terms. There's no pI, but there's a ProtParam mimicking function in BioPython for that. $\endgroup$– Matteo FerlaDec 13, 2021 at 17:49