2:06 PM
1
Based on your recent comment to want to assemble a protein from scratch, and assuming you want to stick to the naturally occurring $\alpha$-amino acids, I suggest to type the amino acids by their three-letter code. ChemDraw has a name2structure conversion routine understanding these codes. If y...
If the properties strongly depend on the 3D structure of the underlying molecules, and not «just» the connectivity of atoms (the .mol files ChemDraw exports while drawing are flat 2D objects), then use, e.g. RDKit or OpenBabel
Then, all hydrogens are in place, and you have access to a big number of molecules with slightly different rotation around the single bonds.
(Do you know what conformers are? I recently encountered a similar question, and the OP was not sure about conformers.)
The point is, you generate these conformers systematically (keyword Ramachandran plot) and sort them by energy.
If you have the hunch an amino acid is missing in your underlying sequence, it likely is easier to weld this missing one into the sequence and see if this still fits within the interpolation of properties.
2 hours later…
4:44 PM
Related, though I haven't had a chance to test it: stackoverflow.com/q/14744216/7696914. FASTA files are used for storing peptide/nucleotide single letter sequences. — Tyberius ♦ yesterday
@neversaint What is the origin of the SMILES string? Given
.pdb
may include information about the amino acid the currently described atom belongs to (fourth column in the atom block here, it seems a well written .pdb
file could be advantageous over a SMILES string (example script, example online service). (No affiliation with either program.) — Buttonwood 4 hours ago@Buttonwood Thank you for your reply. The idea is that I would design the molecule using this PubChem Sketcher. After you draw it, it will give you SMILES. Then from SMILES I would like to make amino acid sequence. — neversaint 4 hours ago
You may type e.g.,
Ala-Cys-Lys
as a string in the name2structure function of Chemdraw (sample page); this draws you the molecule on (electronic) paper. This may be selected (get structure -> get smiles) to generate the SMILES string. Much less hassle than drawing the amino acids atom by atom, and the string written below the drawing below may be exported for a Python-moderated (dictionary based) string substitution into Fasta. With this you tinker only with amino acids known to the program; thus, less chance for errors. — Buttonwood 3 hours agoThe sample page does not know about FASTA, so you have to type the three-letter codes for the amino acid in question. The key space of 21 natural amino acids may be plenty enough. — Buttonwood 3 hours ago
@Buttonwood Thanks so much. It's very useful tool indeed. But the idea is that I have a set of molecular fingerprints which I predict having bioactivities, using machine learning approach. Based on these Fps and using your ChemDraw I'd like to draw then get the peptide (amino acid). Is that a reasonable strategy for novel drug design? Please advice — neversaint 3 hours ago
6 hours later…
10:46 PM
« first day (551 days earlier) ← previous day next day → last day (902 days later) »
Transcript for
Oct24
Nov '211
Nov2
Modeling Matters!
Matters related to modeling matter, because modeling matters!