Updating index splitting number

# DeepChem

# DeepChem

DeepChem aims to provide a high quality open-source toolchain that

democratizes the use of deep-learning in drug discovery, materials science, and quantum

### Performances

Index splitting

|Dataset    |Model               |Train score/ROC-AUC|Valid score/ROC-AUC|Time(loading)/s |Time(running)/s|

|-----------|--------------------|-------------------|-------------------|----------------|---------------| 

|tox21      |logistic regression |0.910              |0.759              |30              |60             |

|           |tensorflow(MT-NN)   |0.987              |0.800              |30              |60             |

|           |robust MT-NN        |0.979              |0.741              |30              |90             |

|           |graph convolution   |0.930              |0.819              |40              |160            |

|muv        |logistic regression |0.910              |0.744              |600             |450            |

|           |tensorflow(MT-NN)   |0.980              |0.710              |600             |400            |

|           |robust MT-NN        |0.986              |0.672              |600             |550            |

|           |graph convolution   |0.881              |0.832              |800             |1800           |

|pcba       |logistic regression |0.794        	     |0.762              |1800            |10000          |

|           |tensorflow(MT-NN)	|0.949        	     |0.791              |1800            |9000           |

|           |graph convolution   |0.866        	     |0.836              |2200            |14000          |

|sider      |logistic regression |0.900        	     |0.620              |15              |80             |

|           |tensorflow(MT-NN)	|0.931        	     |0.647              |15              |75             |

|           |graph convolution   |0.845        	     |0.646              |20              |50             |

|toxcast    |logistic regression |0.762        	     |0.622              |80              |2600           |

|           |tensorflow(MT-NN)	|0.926        	     |0.705              |80              |2300           |

|           |graph convolution   |0.906        	     |0.725              |80              |900            |

Random splitting(Time omitted)

|Dataset    |Model               |Train score/ROC-AUC|Valid score/ROC-AUC|

|-----------|--------------------|-------------------|-------------------|

|tox21      |logistic regression |0.903              |0.705              |

|           |tensorflow(MT-NN)   |0.856              |0.763              |

|           |robust MT-NN        |0.857              |0.767              |

|           |graph convolution   |0.872              |0.798              |

|muv        |logistic regression |0.963              |0.766              |

|           |tensorflow(MT-NN)   |0.904              |0.764              |

|           |robust MT-NN        |0.934              |0.781              |

|           |graph convolution   |0.840              |0.823              |

|pcba       |logistic regression |0.809              |0.776              |

|           |tensorflow(MT-NN)   |0.826              |0.802              |

|           |robust MT-NN        |0.809              |0.783              |

|           |graph convolution   |0.876              |0.852              |

|sider      |logistic regression |0.933              |0.620              |

|           |tensorflow(MT-NN)   |0.775              |0.634              |

|           |robust MT-NN        |0.803              |0.632              |

|           |graph convolution   |0.708              |0.594              |

|toxcast    |logistic regression |0.721              |0.575              |

|           |tensorflow(MT-NN)   |0.830              |0.678              |

|           |robust MT-NN        |0.825              |0.680              |

|           |graph convolution   |0.821              |0.720              |

Random splitting

|Dataset    |Model               |Train score/ROC-AUC|Valid score/ROC-AUC|

|-----------|--------------------|-------------------|-------------------|

|           |robust MT-NN        |0.814              |0.692              |

|           |graph convolution   |0.820        	     |0.692              |

Scaffold splitting(Time omitted)

Scaffold splitting

|Dataset    |Model               |Train score/ROC-AUC|Valid score/ROC-AUC|

|-----------|--------------------|-------------------|-------------------|

|sider      |27         |1427       |

|toxcast    |617        |8615       |

Time needed for benchmark test(~20h in total)

|Dataset    |Model               |Time(loading)/s |Time(running)/s|

|-----------|--------------------|----------------|---------------| 

|tox21      |logistic regression |30              |60             |

|           |tensorflow(MT-NN)   |30              |60             |

|           |robust MT-NN        |30              |90             |

|           |graph convolution   |40              |160            |

|muv        |logistic regression |600             |450            |

|           |tensorflow(MT-NN)   |600             |400            |

|           |robust MT-NN        |600             |550            |

|           |graph convolution   |800             |1800           |

|pcba       |logistic regression |1800            |10000          |

|           |tensorflow(MT-NN)	 |1800            |9000           |

|           |robust MT-NN        |1800            |14000          |

|           |graph convolution   |2200            |14000          |

|sider      |logistic regression |15              |80             |

|           |tensorflow(MT-NN)	 |15              |75             |

|           |robust MT-NN        |15              |150            |

|           |graph convolution   |20              |50             |

|toxcast    |logistic regression |80              |2600           |

|           |tensorflow(MT-NN)   |80              |2300           |

|           |robust MT-NN        |80              |4000           |

|           |graph convolution   |80              |900            |

## Contributing to DeepChem

We actively encourage community contributions to DeepChem. The first place to start getting involved is by running our examples locally. Afterwards, we encourage contributors to give a shot to improving our documentation. While we take effort to provide good docs, there's plenty of room for improvement. All docs are hosted on Github, either in this `README.md` file, or in the `docs/` directory.

on datasets: muv, nci, pcba, tox21, sider, toxcast

time estimation(on a nvidia tesla K20 GPU):

tox21   - dataloading: 30s

        - tf: 40s

muv     - dataloading: 400s

        - tf: 250s

pcba    - dataloading: 30min

        - tf: 2h

sider   - dataloading: 10s

        - tf: 60s

toxcast - dataloading: 70s

        - tf: 40min

(will include more)

Total time of running a benchmark test: 30h

time estimation listed in README file

Total time of running a benchmark test(for one splitting function): 20h

"""

from __future__ import print_function

from __future__ import division