A calculation method that uses protein sequence information to predict protein interaction.

Protein interaction research can reveal the function of proteins at the molecular level and help to reveal the laws of cell activity such as growth and development, metabolism, differentiation and apoptosis.
the identification of protein interactions throughout the genome is an important step in explaining the mechanism of cell regulation.
with the development of protein interaction experimental technology, people can obtain a large number of protein interaction data, and even can analyze protein interaction throughout the genome.
however, due to the limitations of experimental techniques, many high-volume experimental methods have a relatively high error rate of protein interaction data.
, traditional experimental methods are not suitable for detecting large-scale data.
In response to this scientific problem, Wang Yanbin, a master's student in the Multilingeries Information Technology Research Office of Xinjiang Institute of Science and Technology of the Chinese Academy of Sciences, under the guidance of researcher Yu Zhihong, has proposed a method of calculating protein interaction using protein sequence information.
to obtain important protein information, the researchers first used a location-score matrix (PSSM) to represent each protein sequence.
, it is found that the method of representing the quality matrix not only retains the location information of the sequence, but also retains the chemical information of the protein.
at the same time, in order to develop the PCVMZM prediction model, the researchers first extracted accurate and representative protein information from PSSM quality matrix at different scales, and represented each information as a characteristic vector as a feature, using a strong classifier to predict protein interaction.
results show that this method can provide accurate, stable and high coverage prediction information and provide a useful decision-making tool for genomics research.
the study, published in the International Journal of Molecular Science, has been cited 23 times in about two months.
the above research results, researchers have built a deep learning system to achieve a more accurate and stable prediction system.
results show that the prediction accuracy can be improved by 2.2% with the use of deep learning method, and cross-species detection can be achieved.
the study was published in Molecular Biosystems.
work has been funded by the National Natural Science Foundation of China and the "100-person Program" of the Chinese Academy of Sciences.
.