ACCBN: ant-Colony-clustering-based bipartite network method for predicting long non-coding RNA–protein interactions


There have been many studies on protein interaction at home and abroad [27]. There are also many websites which have unveiled a large protein response Database, such as STRING, GEN, BioGRID, DDBJ, Database of Interacting Proteins, ExPasy, Gepasi, etc. [28]. According to the relevant literature, the current studies on protein interaction data are broadly divided into the following three categories:

The first is to determine how proteins interact experimentally. For example, some of the websites mentioned above, such as DIP [29], record the protein data obtained by pure experiments, while other databases [28] also contain the data obtained through experiments. The characteristics of such research results are: the results are true and complete, and the items are complete and functional, but it takes a lot of time, and the preparation of experiments is complicated. However, you get a small amount of data finally. It is impossible to carry out a large number of experiments blindly.

The second is to predict the existence and function of protein interactions with biological theories. This kind of research relies on bioinformatics [27, 30]. Compared with the direct experiments, this kind of method USES some existing data to make predictions. But because there are so many types of protein, there may be a combination of quantity which is very large, the processing efficiency and can deal with the amount of data is still very limited.

The third category is computer algorithms that predict protein interactions. On the basis of the second method, in order to be able to process large data, there are many algorithms for computer prediction interaction [3136]. This method is characterized by large-scale and high efficiency, which can provide more possibilities for the experiments, but since it is a prediction, there will be wrong results. Therefore, three important indicators to test the quality of such methods are computational accuracy, computational efficiency and how much data processed. Because of these advantages of computer methods, more and more researchers are seeking to use better algorithms to predict protein interactions.

The protein interaction network is huge and complex, and the protein reaction confirmed by experiments is only a small part at present. How to expand the known protein interaction network has become a major focus of the research on protein interaction. Biological experiments are time-consuming and expensive, and it is not feasible to test protein pairs one by one. So an effective method commonly used in bioinformatics to expand known protein interaction network rapidly is as follows: first forecast the potential of protein interactions with the known data and then predict the results of the experiment and verify them again.

Our article aimed at developing an efficient and accurate protein prediction method. Only by using the bipartite network prediction algorithm to predict protein interaction, there will be a lot of irrelevant data to reduce the coupling between the data and affect the prediction quality. The ACCBN uses ant colony algorithm to first conduct data clustering, and then constructs a bipartite network for prediction, solving the above problems effectively.

The above experimental results have shown that the prediction results of ACCBN are better than those of other comparison algorithms, and that the prediction results of ACCBN are better than those of other comparison algorithms as well.

Let’s discuss the effect of parameters ρon the prediction accuracy rate results, as shown in Fig. 4.

Fig. 4

The relationship between parameters ρ and prediction accuracy

As can be seen from Fig. 4, as the ρ value increases, the value of the prediction accuracy also increases, but afterρ reaches 0.6, as the ρ value increases, the value of the prediction accuracy begins to decrease. The best prediction accuracy is obtained atρ = 0.6. So we usually set ρ = 0.6 in the experiment.

Articles You May Like

New Math Model Could Impact The Study Of Rapidly Evolving Diseases
How Earth’s mantle is like a Jackson Pollock painting
Dangerous pathogens use this sophisticated machinery to infect hosts
New AI sees like a human, filling in the blanks
Why are gels elastic?

Leave a Reply

Your email address will not be published. Required fields are marked *