RPI-EDLCN

An ensemble deep learning framework based on capsule network for ncRNA–protein interaction prediction

The utils, data and result directories contain model codes, tested data sets and generated results, respectively. The depended python packages are listed in requirements.txt. The package versions should be followed by users in their environments to achieve the supposed performance.

How to run

The program is in Python 3.7.0 using Keras and Tensorflow backends. Use the below bash command to run RPI-EDLCN.

    python main.py -d dataset

The parameter of dataset could be RPI1807,RPI2241 and NPInter v2.0. Then, RPI-EDLCN will perform 5-fold cross validation on the specific dataset.

Three RPI datasets

The widely used RPI benchmark datasets are organized in the data directory.

Due to the limitation of the hardware conditions of the selected RNA secondary structure method, it can only predict the secondary structure of RNA with a length of no more than 1000 nucleotides, so we preprocessed the data.

             Dataset    | #Positive pairs | #Negative pairs  |  RNAs  | Proteins | Reference

Original set

            RPI1807             1807              1436          1078      3131        [1]
			RPI2241             2241              2241          841       2042        [2]
            NPInter v2.0        10412             10412         4636      449         [3]

Optimal set

            RPI1807             652               221           646       868         [1]
			RPI2241             872               872           582       1190        [2]
            NPInter v2.0        3216              3216          1085      449         [3]

Help

For any questions, feel free to contact me by tanjianjun@bjut.edu.cn or start an issue instead.

[1] Pan, X.Y.; Fan, Y.X.; Yan, J.C.; Shen, H.B. IPMiner: hidden ncRNA-protein interaction sequential pattern mining with stacked autoencoder for accurate computational prediction. Bmc Genomics 2016, 17. doi:ARTN 582 10.1186/s12864-016-2931-8.

[2] Peng, C.; Han, S.; Zhang, H.; Li, Y. RPITER: a hierarchical deep learning framework for ncRNA–protein interaction prediction. Int J Mol Sci. 2019;20(5):1070. doi: 10.3390/ijms20051070.

[3] Yuan, J.;Wu,W.; Xie, C.Y.; Zhao, G.G.; Zhao, Y.; Chen, R.S. NPInter v2.0: an updated database of ncRNA interactions. Nucleic Acids Research 2014, 42, D104–D108. doi:10.1093/nar/gkt1057.

Reference:

RPI-EDLCN: an ensemble deep learning framework based on capsule network for ncRNA–protein interaction prediction