Hi,
I am new to deepchem. Currently, I am now try to train a model to calculate the complxation energy of Host-Guest interaction. Therefore, I am trying to find the avalible dataset on the internet. So far, I have collected only around 500 samples... Is there a dataset in the deepchem? Or do you have any suggestion for me to find the datasets? Thank you very much!
I am new to deepchem. Currently, I am now try to train a model to calculate the complxation energy of Host-Guest interaction. Therefore, I am trying to find the avalible dataset on the internet. So far, I have collected only around 500 samples... Is there a dataset in the deepchem? Or do you have any suggestion for me to find the datasets? Thank you very much!
4 replies
@kurokawaikki Could you clarify what you mean by host-guest interaction? Sorry my knowledge of materials science might be lacking here!
i'm try to recreate this paper they use 15 propoties
https://pubs.acs.org/doi/10.1021/acs.jpca.8b09376
https://pubs.acs.org/doi/10.1021/acs.jpca.8b09376
You can find a description of the qm9 labels here: https://deepchem.readthedocs.io/en/latest/api_reference/moleculenet.html#deepchem.molnet.load_qm9
You'll want to pick out the same 15 properties as in that paper
i might do something wrong , maybe my model is really bad that why it not close to the input label data
@atreyamaj That would be a great idea :). If you'd like to help, DM me your pypi username and I'll give you access to edit the description. It might be good to make the description documented in the main repo and add a note to the release docs to update the pypi description if needed
@alat-rights It's a little tricky basically. The flaky tests are already separated out so most of their failures don't affect the CI but we have a hundreds of tests and we run into edge cases that are hard to route around. The only solution right now is to just look into each case individually and try to understand the cause of the failure, but maybe there's a better idea
Hello! I am new to AI drug and deepchem library. I have to say this work is remarkable and very user-friendly to beginners like me. I am now trying to play with the models and reproduce the results in http://moleculenet.ai/latest-results, I find the benchmark scripts in deepchem/examples/benchmark.py, I am wondering whether you happen to keep the record of hyperparameters to reproduce the results ?
@MhDang Welcome to the project! I'd suggest checking out the new moleculenet repo: https://github.com/deepchem/moleculenet
We have a new leaderboard up with maintained results. The original model benchmarks were run on TF 1.x and the underlying libraries have changed a lot so it's not easy to directly replicate those results
You can find the original hyperparameters at https://github.com/deepchem/deepchem/blob/2.3.0/deepchem/molnet/preset_hyper_parameters.py
@rbharath I am sorry for the late responses. In my research, the host compounds are the cyclic (or ring like) compounds such as valinomycin. I try to find a guest compound that can fit into the ring area and predicts the binding energy. Host–guest chemistry is currently being tested in SAMPL challenge. Since I would like to build a prediction model, I hope I could have as many sample as possible. Therefore, is there a dataset in the deepchem? Or do you have any suggestion for me to find the datasets? Thank you very much!
@kurokawaikki Ah, I see! That makes sense. Hmm, unfortunately, I'm not aware of a good dataset in DeepChem for host-guest interactions. Closest would be pdbbind but that's for more generic protein-ligand interactions and not for the types of host-guest interactions you're envisioning
Hi everyone,
I have working with generative modelling for Molecules (SMILES) and I was exploring the AspuruGuzikAutoEncoder given on seqtoseq.py. The original paper has a step for Gaussian Process step for exploring the latent space and I couldn't find its implementation in deepchem. It would be really helpful if someone could suggest me generative models research or frameworks which can provide us with the option of exploring the latent space for finding more optimized molecules.
reference -
- Aspuru Guzik's Mol VAE paper - https://arxiv.org/abs/1610.02415 (Gaussian Process is given in Page 11 , Optimization of molecules via properties)
Thanks in advance :)
I don't think we have a good out-of-box technique for exploring the latent space but something should work
*might work :)
I found this repo for the paper you linked above: https://github.com/HIPS/molecule-autoencoder
It's outside of deepchem, but hope this helps!
@VIGNESHinZONE
_criterionmae_max_depth_8_min_samples_leaf_3_min_samples_split_3_min_weight_fraction_leaf_0_n_estimators_120: 0.6832978488194399 here is the result of the first hyperparameter research
and the r2 score of testing : 0.7238619516613416
and the r2 score of testing : 0.7238619516613416
when i add another value to "min samples leaf" the validation score and r2 score of testing are changing
so the validation scores :
_criterionmae_max_depth_8_min_samples_leaf_1_min_samples_split_3_min_weight_fraction_leaf_0_n_estimators_120_njobs-1: 0.6514039887382881
_criterionmae_max_depth_8_min_samples_leaf_3_min_samples_split_3_min_weight_fraction_leaf_0_n_estimators_120_njobs-1: 0.6667454007419835
and the r2 score of testing : 0.7514118547850513
_criterionmae_max_depth_8_min_samples_leaf_1_min_samples_split_3_min_weight_fraction_leaf_0_n_estimators_120_njobs-1: 0.6514039887382881
_criterionmae_max_depth_8_min_samples_leaf_3_min_samples_split_3_min_weight_fraction_leaf_0_n_estimators_120_njobs-1: 0.6667454007419835
and the r2 score of testing : 0.7514118547850513
Hello guys, I was working on issue deepchem/deepchem#631 and opened up a draft PR deepchem/deepchem#2501. I'm quite new to deepchem and would appreciate any help I can get. It involved the use of the
BindingPocketFeaturizer.
@gokhantahil Sorry, I'm not sure what the bug is here. Would you mind clarifying a bit more?
Hi everyone - I have a basic question about the pdbbind data. My understanding is that a pdbbind model can take either the ligand or the protein-ligand complex as input, and produce -ln(kd/ki) as output. Are the proteins all the same or are they different? If they're different, how can a purely ligand-based model be trained to predict -ln(kd/ki)? Wouldn't it also need some information about the protein as input?
@simonaxelrod The proteins are different. (I think a few proteins are repeated but these are the exceptions). The purely ligand models are really learning a measure of "ligand-ness" and are more of a baseline control on the protein-ligand models. The delta between the protein-ligand model and the ligand-only model are a measure of how much information about the protein the model is actually using
Thanks @rbharath! That makes sense that the ligand models are really just a baseline control. Though I think some papers would benefit from saying this explicitly - for example, the ChemProp paper just notes that their model outperforms all moleculenet models for all tasks other than QM and pdbbind. But should probably have said that it definitely shouldn't work for pdbbind or something would be really wrong
I'm looking into featurizing a set of molecules with the ConvMolFeaturizer. I'm interested in featurizing the chemical environment of the atoms within the molecule so I presume that I'd want to set the per_atom_fragmentation parameter. In the docs it notes:
This option is typically used in combination with a FlatteningTransformer to split the lists into separate samples.
I can't find any mention of FlatteningTransformer in the docs, can someone point me somewhere?
@kingscolour
per_atom_fragmentation is a new feature so this may be a docs error. Check out the new tutorial at https://github.com/deepchem/deepchem/blob/master/examples/tutorials/Training_a_Normalizing_Flow_on_QM9.ipynb
Thanks! I actually missed that one because I didn't skipped over the files without a number. The Atomic Contributions for Molecules tutorial was also helpful for my understanding. Cheers for your work!