Pairwise input neural network for target-ligand interaction prediction

Wang, Caihua; Liu, Juan; Luo, Fei; Tan, Yafang; Deng, Zixin; Hu, Qian-Nan

doi:10.1109/bibm.2014.6999129

Cited by 28 publications

(23 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the capacity to learn higher level of feature representation, deep learning has become an effective and popular tool in many fields, including Bioinformatics community [15]. For example, Wang et al proposed a pairwise input neural network for target-ligand interaction prediction [16]; Aliper et al used deep learning to predict pharmacological properties of drugs based on transcriptomic data [14]; Wan et al also successfully predicted the compound-protein interactions by deep learning [17].…”

Section: Introductionmentioning

confidence: 99%

Pathway-guided deep neural network toward interpretable and predictive modeling of drug sensitivity

Liu

Deng

Cai

et al. 2020

Preprint

View full text Add to dashboard Cite

Motivation: To efficiently save cost and reduce risk in drug research and development, there is a pressing demand to develop in-silico methods to predict drug sensitivity to cancer cells. With the exponentially increasing number of multi-omics data derived from high-throughput techniques, machine learning-based methods have been applied to the prediction of drug sensitivities. However, these methods have drawbacks either in the interpretability of mechanism of drug action or limited performance in modeling drug sensitivity. Results: In this paper, we presented a pathway-guided deep neural network model, referred to as pathDNN, to predict the drug sensitivity to cancer cells. Biological pathways describe a group of molecules in a cell that collaborates to control various biological functions like cell proliferation and death, thereby abnormal function of pathways can result in disease. To make advantage of both the excellent predictive ability of deep neural network and the biological knowledge of pathways, we reshape the canonical DNN structure by incorporating a layer of pathway nodes and their connections to input gene nodes, which makes the DNN model more interpretable and predictive compared to canonical DNN. We have conducted extensive performance evaluations on multiple independent drug sensitivity data sets, and demonstrate that pathDNN significantly outperformed canonical DNN model and seven other classical regression models. Most importantly, we observed remarkable activity decreases of disease-related pathway nodes during forward propagation upon inputs of drug targets, which implicitly corresponds to the inhibition effect of disease-related pathways induced by drug treatment on cancer cells. Our empirical experiments show that pathDNN achieves pharmacological interpretability and predictive ability in modeling drug sensitivity to cancer cells. Availability: The web server, as well as the processed data sets and source codes for reproducing our work, is available at

show abstract

Section: Introductionmentioning

confidence: 99%

Pathway-guided deep neural network toward interpretable and predictive modeling of drug sensitivity

Liu

Deng

Cai

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…However, the advantage of DNN is to learn a be er feature representation, but it fails to consider the essential imbalance issue in friend recommendation. While a few e orts have been made using BPR (Bayesian Personalized Ranking) or DNN (Deep Neural Network) [41] for friend recommendation, no work has been done combining the strength of these two ideas.…”

Section: Previous Work On Deep Neural Network (Dnn)mentioning

confidence: 99%

“…Another reason is that DNN usually needs proper pre-training strategy to avoid poor parameter initialization, e.g., use pre-trained deep neural network like VGG-Net to initialize parameters of model [24,32] However, it is unclear which pre-training strategy should be used for DNN in friend recommendation. Wang et al [41] tried to use a four-layer neural network with a simple pre-training strategy called Pairwise Input Neural Network (PINN) to do link prediction, by training and predicting the probability of a link point-wisely. However, it did not achieve signi cant performance improvement.…”

Section: Related Workmentioning

confidence: 99%

“…Pairwise Input Neural Network (PINN) [41]: It is a fourlayer arti cial neural network with simple pre-training strategy for link prediction, which regarded two nodes' link vector as input and output the probability that there exists a link between them.…”

Section: Experiments 41 Experiments Settingsmentioning

confidence: 99%

“…Update BayDNN with AdaDelta 19: end procedure Note that previous work [41] has a similar two-step pre-training design for a di erent application. However, for the rst subproblem, it directly obtains the value for F * with a greedy algorithm.…”

mentioning

confidence: 99%

See 2 more Smart Citations

BayDNN

Ding

et al. 2017

Proceedings of the 2017 ACM on Conference on Information and Knowledge Management

View full text Add to dashboard Cite

Friend recommendation is a critical task in social networks. In this paper, we propose a Bayesian Personalized Ranking Deep Neural Network (BayDNN) model for friend recommendation in social networks. BayDNN rst extracts latent structural pa erns from the input network data and then use the Bayesian ranking to make friend recommendations. With BayDNN we achieve signi cant performance improvement on two public datasets: Epinions and Slashdot. For example, on Epinions dataset, BayDNN signi cantly outperforms the state-of-the-art algorithms, with a 5% improvement on NDCG over the best baseline. e advantages of the proposed BayDNN mainly come from a novel Bayesian personalized ranking (BPR) idea, which precisely captures the users' personal bias based on the extracted deep features, and its underlying convolutional neural network (CNN), which o ers a mechanism to extract latent deep structural feature representations of the complicated network data. To get good parameter estimation for the neural network, we present a netuned pre-training strategy for the proposed BayDNN model based on Poisson and Bernoulli probabilistic models.

show abstract

Applications of Deep Learning in Healthcare and Biomedicine

Mittal

Hasija

2019

Studies in Big Data

View full text Add to dashboard Cite

Pairwise input neural network for target-ligand interaction prediction

Cited by 28 publications

References 17 publications

Pathway-guided deep neural network toward interpretable and predictive modeling of drug sensitivity

Pathway-guided deep neural network toward interpretable and predictive modeling of drug sensitivity

BayDNN

Applications of Deep Learning in Healthcare and Biomedicine

Contact Info

Product

Resources

About