Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Vignesh Ram Somnath
    @VIGS25
    @CreamyLong Number of filters in the Convolutional Layer
    Qian Long
    @CreamyLong
    @VIGS25 Thank you
    Bharath Ramsundar
    @rbharath
    @MakiNakahara It looks like you're on a very old version of DeepChem (1.3 I think). Would it be possible for you to try upgrading to a more recent version?
    MakiNakahara
    @MakiNakahara
    @rbharath I use Deepchem 2.1.1.dev353 . When I try upgrading to most recent version (2.2.1.dev54), dc.nn.SequentialGraph and DTNNMultitaskGraphRegressor is removed and I can't find similar Class and Function. How do you
    @rbharath What Class and Function do you use when you use deepchem to custom the number of layer and the type of layer?
    Bharath Ramsundar
    @rbharath
    Check out some of the tutorials on graph convolutions
    MakiNakahara
    @MakiNakahara
    @rbharath Thank you. I'm trying to check the source code.
    Jeraime Griffith
    @jgriffi

    Hi I'm getting started with deepchem and setting up an environment following the text below:

    git clone https://github.com/deepchem/deepchem.git
    cd deepchem
    bash scripts/install_deepchem_conda.sh deepchem
    source activate deepchem
    python setup.py install
    nosetests -a '!slow' -v deepchem --nologcapture

    And now I'm getting this message:
    libc++abi.dylib: terminating with uncaught exception of type std::__1::system_error: thread::join failed: No such process

    Any ideas what's wrong?

    Bharath Ramsundar
    @rbharath
    Hmm, this could be an rdkit error
    Or possibly from tensorflow. These are the two C++ packages
    Can you see if you can get these packages installed independently
    Ghost
    @ghost~5d3c94cad73408ce4fc71b58
    any help ? please, thanks!
    Jeraime Griffith
    @jgriffi
    @rbharath I don't underastand. What packages should I install indepoendently?
    Vignesh Ram Somnath
    @VIGS25

    @jgriffi TensorFlow and RDKit.

    Commands:

    conda install -c conda-forge rdkit
    pip install tensorflow==1.13.1

    Use tensorflow-gpu instead of tensorflow if you want to install the GPU version

    RishikeshMagar
    @RishikeshMagar
    Hi,
    I have a dataset where I have two input molecules. I plan to featurize them separately and combine the features and subsequently pass it to GraphConvModel class. The tutorials that I have seen using Graph Convolutions are with a single molecule as input. Any help with proceeding on these type of problems would be appreciated
    Vishesh Mangla
    @XtremeGood
    hi
    someone here?
    Vishesh Mangla
    @XtremeGood
    I need to find the coupling constant from the given data of muliken charge, xoordinates of molecules in .xyz files , potential energy etc.
    How should I began to think
    I 'm a chemistry student and know the basics of nmr and proton shielding stuff.
    Sabrina Ho
    @iamsabhoho
    hi, im doing a research on molecule similarity search. please check out my blog: https://medium.com/gsi-technology/rdkit-for-newbies-3697e617521f
    Vignesh Ram Somnath
    @VIGS25
    @iamsabhoho: You could also consider posting it here: https://forum.deepchem.io/
    Sabrina Ho
    @iamsabhoho
    @VIGS25 ok thank!
    Sabrina Ho
    @iamsabhoho
    hi, i just finished the second blog on my molecule similarity search. This time I transformed them into vectors! https://medium.com/gsi-technology/tanimoto-vs-mol2vec-7fa4af3208ef
    HL
    @HL22233092_twitter
    Hello everyone, I'm trying to install deepchem. I tried different anaconda versions and the commands, but none of them worked...
    Then I tried to install from source in a conda environment. The first few steps went well until I ran nosetests -a '!slow' -v deepchem --nologcapture.
    The error is Tests that scores are generated ... libc++abi.dylib: terminating with uncaught exception of type std::__1::system_error: thread::join failed: No such process
    It kept coming out, so I have to terminate it by control + c.
    HL
    @HL22233092_twitter
    I can still use the deepchem env in jupyter notebook, so does this error really mater?
    Mostafa Nategh
    @1particle
    Facing the same issue here, except that I haven't tried installing from source. None of the two easy installation methods listed on the git repo (pip3 and conda) seem to be fully working on Ubuntu 18.04 right now: pip3 method works, but it won't give you rdkit (with python 3), which is not really useful. Conda either won't complete installation or installs, but then python (3.x) cannot find deepchem. I hope there's an easy fix to this.
    Mostafa Nategh
    @1particle
    Installing with python 2.7, the conda way, also failed on a fresh install of ubuntu 18.04. I'm very sad right now!
    Vignesh Ram Somnath
    @VIGS25
    This is very strange. I'm going to try and work on this.
    catenate15
    @catenate15
    @VIGS25 I am also facing similar problems installing deepchem . I have tried installing via anaconda both on windows and mac os but it still failed. I hope this can be fixed soon.
    Anup Adhikari
    @anupadkh
    @HL22233092_twitter , @1particle The solution to it is before importing rdkit: import sys; sys.path.append(' .....rd-kit_path....'); Also make sure you export $PATH to include the path of rdkit installed bin path. (aka Environment Variables in Windows). I hope this helps
    catenate15
    @catenate15
    @anupadkh , please could you explain explicitly how I could go about this using anaconda. I am quite new to programming. Thanks
    Usman Zafar
    @OsmanXafar_twitter
    Is there a way to feed your own dataset (Smiles,Labels) into deepchem for regression tasks?
    jordenkass
    @jordenkass
    @catenate15 Try the installation guide I put on github
    catenate15
    @catenate15
    Thanks @jordenkass .
    catenate15
    @catenate15
    @jordenkass , I tried installing it on linux (kubuntu) but still got this error.
    image.png
    using Spyder (Anaconda), this is the error message I received.
    image.png
    jordenkass
    @jordenkass
    I didn't install deepchem with conda. I installed directly from the repository as I said in my tutorial. What version of python are you using? I don't think it works right with 3.7
    catenate15
    @catenate15
    Guess I will install python 2.7 then
    gdreiman1
    @gdreiman1
    Is there a reference for the GraphConvModel default architecture? The 2016 Molecule Net paper references the Low Data Drug Discovery paper, but the architecture in the appendix there seems quite different from the current defaults.
    HL
    @HL22233092_twitter
    Hello everyone, I'm working on the Case Study: Training an MNIST Model from the book. After I have finished the training , and the cost is pretty low, but the accuracy is not good. Do you know why that is the case? Thank you!
    Screen Shot 2019-08-10 at 11.10.41 PM.png
    Goekman Altay
    @altayg
    Screenshot from 2019-08-21 14-31-57.png I got the error message in the picture while running this tutorial: https://deepchem.io/docs/notebooks/dataset_preparation.html . It looks the error is caused while working on this drug: 'Colistimethate┬ĚNa'. Anyone had the same issue while running this tutorial or is there a solution suggestion for this? Thanks