Detect The Insecticide Resistance And Invasiveness Of Insects With Omics Data

Agricultural pests and invaded insects have caused huge losses to global agriculture every year.In recent years,due to excessive use of pesticides,many pests have produced high levels of resistance,and became a huge threat to agricultural production and human health.At the same time,our country is also in danger of invaded insects because of its vast territory,long international line length,frequent international trade and so forth.Therefore,the study of insect resistance and invasiveness is of great significance to the sustainable development of agriculture,the ecological environment and the human health.In this paper,we use bioinformatics and machine learning techniques to study insect resistance and invasiveness.(1)Acetylcholinesterase is the target of organophosphorus pesticides and pyrethroid pesticides which are two most widely used pesticides.The gene mutation of acetylcholinesterase,which makes it insensitivity to insecticides,is an important cause of insect resistance.In this thesis,we directly use transcriptome to detect the mutation sites of acetylcholinesterase which relate to pesticides resistance.It can shorten the time needed to detect the mutation sites using the traditional PCR method which need huge time.A database of the total acetylcholinesterase genes and the gene mutations related to resistance were established.We build a pipeline to find the resistant reads and sensitive reads,which use Bowtie2 for rapid short sequence alignment..The resistance frequency can also be calculated by our pipeline.A website was established for the algorithm,and the online system is provided to facilitate the other researchers.(2)The reason for the success of insect invasion is an important problem for the ecologists.The main methods to predict invasive insect invasion risk are based on the phenotype of insects and environmental effects.At present,a large number of insect genomes have been sequenced but not been fully effectively analyzed.In this paper,we use genomics data and machine learning technology to predict the invasiveness of insects.In this paper,gene family data was extracted from 30 genomes,including invasive insects and non-invasive insects.A feature selection algorithm based on multi population genetic algorithm and a feature selection algorithm based on discrete degree of test set SVM-RFE are designed.Different types of classifiers are applied to predict insect invasiveness.At the same time,the feature selection algorithm is compared with the 3 existing feature selection algorithms.We achieve a model which has a strong ability to predict the probability of invasion,and acquire 95%accuracy with 10 fold cross validation.