On Legal Protection Technology Research of Personal Information Based on Big Data

Fei Teng

Abstract: With the arrival of the era of big data, the life of people ushered a new change. At the same time of using big data technology to provide convenience for the social life of people, we must see that big data technology is prone to cause the problem of personal information disclosure, which can results in the invasion of the privacy space of the public. This paper starts with the definition of personal information right in the era of big data, analyzing the personal information risks that may be caused by big data technology, based on this, the protection technology is discussed at the legal level.

: Download (free)  CSE-PDF

QSAR study on potent inhibitor of the menin-mixed lineage leukemia protein-protein interaction based on multi-kernel SVR

Junkang Liu, Mingyu Zhang, Deshun Bi, Jize Wu1, Honglin Zhai

Abstract: Research shows that inhibition of the menin-mixed lineage leukemia protein-protein interaction is a promising new therapeutic strategy for the treatment of acute leukemia carrying MLL fusion (MLL leukemia). My work is to evaluate the properties of the compounds that can inhibit the menin-mixed lineage leukemia protein-protein interaction with a multi kernel SVR model. The prediction models were established based on the IC50 of compounds. The correlation coefficient (R2) was 0.7165 in the best linear model of IC50. The R2, MSE, MSEcv of train set with SVR model are 0.982, 0.128, 0.175, the R2, MSE of test set are 0.938 and 0.175, respectively.
In this paper, improved radial basis function kernel and polynomial kernel are chosen to construct a multi kernel function. The particle swarm optimization algorithm is used to optimize kernel function coefficients and parameters to verify the effectiveness in QSAR modeling. The best satisfied model will provide reference for the design of new potent inhibitor.

: Download (free)  CSE-PDF

Predicting the activity of novel anaplastic lymphoma kinase by machine learning

Wenhui Ren, Aili Qu, Peijian Zhang, Honglin Zhai

Abstract: Anaplastic lymphoma kinase has been turned out to be associated with many human neoplasms, which makes it an attractive therapeutic target, especially in the treatment of non-small cell lung cancer. So many related researches have been conducted to find new anaplastic lymphoma kinase drugs with higher activity. In this paper, four quantitative structure-activity relationship models were established on the base of a series of calculations and analysis of 70 novel synthetic compounds targeted anaplastic lymphoma kinase. One linear model was built up by heuristic method and three non-linear models were separately built by gene expression programming, support vector machine, and gradient boosting decision tree. Comparing the prediction performance of these models through the square of correlation coefficient and the mean square error, gradient boosting decision tree shows the best stability and predictive ability on both training and testing sets. which will be of great help in the design of anaplastic lymphoma kinase inhibitors. The nonlinear regression model based on gradient boosting decision tree will be of great help in the design of anaplastic lymphoma kinase inhibitors.

: Download (free)  CSE-PDF