Functional annotation of creeping bentgrass protein sequences based on convolutional neural network


الملخص بالإنكليزية

Background: Creeping bentgrass (Agrostis soionifera) is a perennial grass of Gramineae, belonging to cold season turfgrass, but has shallow adventitious roots, poor disease-resistance. Little is known about the ISR mechanism of turfgrass and the signal transduction involved in disease-resistance induction, especially the function of a large number of disease-resistance related proteins are urgent to be explored. Results: In this work, the protein sequences of creeping bentgrass were measured and annotated by a functional prediction model based on convolutional neural network. Creeping bentgrass seedlings were grown with BDO treatment, and the ISR response was induced by infecting Rhizoctonia solani. We preformed the transcriptome analysis by Illumina Sequencing and high-quality unigenes were obtained. A minority of assembled unigenes were functionally annotated according to the database alignment while a large part of the obtained amino acid sequences was left non-annotated. To treat the non-annotated sequences, a prediction model was established by training the data set from GO families in three domains to acquire good performance, especially the higher false positive control rate. With such model, we analyzed the non-annotated protein sequences of creeping bentgrass transcriptome, and annotated the disease-resistance response and signal transduction related proteins. Conclusions: The results provide good candidates of the proteins with certain functions. With the results in this work, the waste of transcriptome sequencing data of creeping bentgrass can be avoided, and research time and labor for the analysis of ISR characteristics of creeping bentgrass will be saved in further research. It also provides reference for the sequence analysis of turfgrass disease-resistance research.

تحميل البحث