Problem with using aromatic rings as starting compounds

jdurrant — Thu, 12 Sep 2019 05:31:24 -0000

Hi Patrick. Much thanks for your interest in AutoGrow, and please forgive my delay in getting back to you. We're currently putting together a new version of AutoGrow with some major improvements. My student Jake Spiegel has rewritten and modernized almost the entire codebase. We hope to submit the work for publication in a few months. I'm not sure why exactly the old version isn't processing your input molecules properly, but I'm confident the new version will be able to handle them if they are amenable to one of the included reaction schemes. Shouldn't be too much longer until publication. Thanks again for your interest. All the best.

Problem with using aromatic rings as starting compounds

jdurrant — Thu, 12 Sep 2019 05:31:22 -0000

Problem with using aromatic rings as starting compounds

patrick connolly — Sun, 04 Aug 2019 12:05:10 -0000

Hi jd/AutoGrow support,

I'd like to start out by saying I like the tool, it works pretty well, however I have encountered a strange problem. I have been trying to replicate the first procedure outlined in the paper. In this, you used a panel of all possible permutations of brominated benzene as starting compounds.

After a ton of trial and error, I think I at least know the source of the problem. My process for PDB generation is i) download the SD file from Pubchem and ii) convert it to PDB with openbabel. When I try this with other compounds or fragments, it works perfectly, and I can get to the vina screening stage.

However, when I try to use any of the brominated benzene compounds, the program always hangs on the initial "creating new ligands" step. If I change the bromines to either chlorines or fluorines, I get the same result. Same if I use simple benzene as a starting compound. My theory is that the program is trying to calculate bond angles, but because it is a simple cyclical molecule, the program ends up going around and around in a circle, unable to reach a conclusion.

Do you guys have any insight into how to get over this? How did you do it in the paper example?

Best wishes,
Patrick

Bacterial Cell Filter?

jdurrant — Thu, 28 Sep 2017 13:41:11 -0000

Hi Andrew. Please forgive my ridiculous delay in getting back to you. I wanted to let you know that we have begun to work on Autogrow 4. We intend to make it more modular so that custom filters can be easily added. I expect we won't be ready to publish for some months, but I wanted to let you know we're working on it! All the best.

Bacterial Cell Filter?

Andrew Hogan — Fri, 30 Jun 2017 16:53:10 -0000

Hi there,
I was wondering if in the future you had plans for developing a filter for (Gram negative) bacterial cells? Its easy to find in vitro enzyme inhibitors, but very few compounds are capable of crossing the bacterial cell envelope, especially the Gram-negative cell envelope. My lab has an antibiotic development project that I've been working on for a while now. I've been using autogrow successfully for a few weeks, but my supervisor isn't as happy with the results as I expected. My supervisor would be much more willing to follow up on the results if we had some indication that the ligands could cross the envelope.

Thanks very much
Andrew

Marking ligand atoms

jdurrant — Sun, 21 May 2017 05:02:58 -0000

Hi Francesco. Did you end up getting the run to work?

Making crossover issues

jdurrant — Sat, 20 May 2017 17:10:23 -0000

Hi Francesco. If I'm not mistaken, autogrow can filter my molecular weight. That's one of Lipinski's rules.

Marking ligand atoms

Francesco Pietra — Sun, 07 May 2017 14:07:16 -0000

I now noticed that those infos about deleting ligands are contained in log.txt, and where present also in previous autogrow sessions with the two families of initial ligands.

I forgot to add the input:

-filename_of_receptor 5000_less_TA1_Mg.pdb \ -center_x 325.522 -center_y 448.689 -center_z 344.046 \ -size_x 88 -size_y 94 -size_z 126 \ -additional_autoclickchem_parameters "all_reactions" \ -allow_modification_without_frag_addition TRUE \ -directory_of_source_compounds ./starting_sarco/ \ -directory_of_fragments ../fragments/MW_200/ \ -number_of_mutants_first_generation 10 -number_of_crossovers_first_generation 10 \ -number_of_mutants 10 -number_of_crossovers 10 \ -top_ones_to_advance_to_next_generation 5 -num_generations 2 \ -max_seconds_per_generation 1000000 \ -use_lipinski_filter FALSE -use_strict_lipinski_filter FALSE -use_ghose_filter FALSE \ -scoring_function VINA -score_by_ligand_efficiency FALSE \ -maintain_core TRUE -minimum_core_atoms_required 19 \ -vina_executable /opt/vina_1_1_2_linux_x86/bin/vina \ -openbabel_bin_directory /usr/bin/ \ -mgltools_directory /opt/MGLTools-1.5.6/ \ -num_processors 8 \ -output_dir ./autogrow_new_out/

 I was wondering whether i should rather start from the core structure. However, I allowed also refuncionalization (-allow_modification_without_frag_addition TRUE \), which should rapidly generate new ligands with all 19 marked atoms.

Marking ligand atoms

Francesco Pietra — Sun, 07 May 2017 13:06:26 -0000

I have started a new autogrow session with half of previous starting structures, i.e., those of one more homogeneous family only of structures.

On ctrl-C, killing the execution and then restarting with same settings, I noticed a series of infos like thais one:

Deleting ligand /home/francesco/work_autogrow/sarco_tubulin/autogrow_sarco_tub_out/generation1/157752.mutant.pdb because it contains only 10 marked ("core") atoms.

I should inform that I wanted to preserve the core structure of this family of ligands (19 atoms out of the 72 atoms composing the whole molecule), appending a "!" to the atom name of the atoms cmposing the core structure. In vie of the difficulty of getting ligands with the 19 aoms marke, I wonder whether this can be an acceptable input.

I woud also like to know where getting the above infos during excecution and not only when killing the execution

thanks for advice.

Making crossover issues

Francesco Pietra — Sat, 06 May 2017 09:14:46 -0000

Hi Jacob: The "dimeric" crossovers are indicated in /support by "number.crossover.pdb.can"

As a side chain of the involved starting structure ends with a monosaccharide unit, propagation ("dimerization") occurs there at one free -OH. That is, the genetic algorythm carries out its job in the absence of a filter as to the MW.

Is that possible to modify the .py code by inserting an upper limit to the MW? Or does it exist in the literature a Lipinski-like filter for high MW?

thanks

Recent posts to Discussion

Problem with using aromatic rings as starting compounds

Problem with using aromatic rings as starting compounds

Problem with using aromatic rings as starting compounds

Bacterial Cell Filter?

Bacterial Cell Filter?

Marking ligand atoms

Making crossover issues

Marking ligand atoms

Marking ligand atoms

Making crossover issues