A method for cancer marker screening based on multi-task sample selection

Abstract

The application discloses a cancer marker screening method based on multi-task sample selection, comprising the following steps: step 1, collecting cancer gene expression data sets and performing data preprocessing; step 2, calculating the instance importance of samples to construct an auxiliary task data set; step 3, initializing a main task population and an auxiliary task population based on the preprocessed data; step 4, respectively calculating the optimization target value of each individual in the two task populations, and generating offspring by adopting different evolution strategies; step 5, performing knowledge transfer between the two tasks every k generations, and reconstructing the auxiliary task every r generations; step 6, iterative optimization, and selecting the best sample subset; and step 7, inputting the sample subset into a feature selection algorithm to obtain the final gene expression combination, that is, screening out suspected cancer markers. The application can select the best sample subset, thereby improving the time efficiency of cancer marker screening, and screening out better cancer markers.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and machine learning, specifically a method for screening cancer markers based on multi-task sample selection. Background Technology

[0002] In recent years, with the advancement of high-throughput sequencing technology, RNA cancer gene expression profiling has become an important resource for elucidating disease mechanisms and screening novel biomarkers. Changes in RNA expression levels not only reflect the dynamic regulation of gene activity but are also closely related to cellular physiological states and disease progression. Analyzing the close relationship between genes and cancer, and exploring the selection of the most representative and predictive genomic collaborations from massive RNA expression data as potential cancer biomarkers to enhance predictive ability and clinical applicability, has become a current research hotspot.

[0003] Existing methods for screening potential cancer markers from RNA cancer gene expression profiles using feature selection algorithms have significant limitations.

[0004] 1. Current screening methods do not consider the quality of the dataset. Technical limitations of sequencing instruments and incomplete sequencing chemical reactions can generate noise and erroneous data, which can affect the screening results for cancer markers.

[0005] 2. Each sample contains tens of thousands of genes, resulting in a large number of redundant samples in the dataset. Processing this data will consume a significant amount of time and affect the efficiency of cancer marker screening. Summary of the Invention

[0006] The present invention aims to address the shortcomings of the existing technology by proposing a cancer marker screening method based on multi-task sample selection. This method first aims to reduce and optimize the sample dataset, thereby improving the effectiveness and efficiency of cancer marker screening.

[0007] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0008] The cancer marker screening method based on multi-task sample selection of the present invention is characterized by the following steps:

[0009] Step 1: Obtain the cancer gene expression dataset {(x w ,y w )|w=1,2,…,W}, where, (x w ,y w ) represents the w-th cancer gene expression sample, x w Let w be the set of attribute features of the w-th sample, and Let y represent the d-th attribute feature of the w-th sample, where d∈[1,D], and D is the total number of attribute features of the cancer gene expression samples. w Let y represent the class label of the w-th sample. w w∈[1,C], where C represents the total number of categories; w∈[1,W], where W is the total number of cancer gene expression samples;

[0010] Step 2: Construct the auxiliary task dataset;

[0011] Step 2.1, from {(x w ,y w Obtain all samples of the c-th class from the set |w = 1, 2, ..., W}, and calculate the average of the attribute features of all samples of the c-th class, which is used as the center vector of the c-th class. c ; Calculate the expression of the w-th cancer gene sample and the center c After calculating the Euclidean distance and normalizing it, the importance imp of the w-th sample is obtained. w Thus, the importance of all samples is obtained: imp = {imp} w |w=1,2,…,W};

[0012] Step 2.2: Define the algebra of the auxiliary task dataset as t, and initialize t = 1;

[0013] Define the t-th generation flag vector in, This represents the w-th flag in the flag vector of generation t;

[0014] Construct a random probability vector pos = {pos w |w=1,2,…,W};where, pos w This represents the w-th random number;

[0015] If pos w <imp w Then let This indicates that the w-th cancer gene expression sample (x) is retained. w ,y w Otherwise, let This indicates the deletion of the w-th cancer gene expression sample (x). w ,y w This allows for the construction of a generation t auxiliary dataset for cancer gene expression. in, This represents the auxiliary sample for the expression of the u-th cancer gene in generation t; Let represent the set of attribute features of the u-th auxiliary sample in the t-th generation. U represents the class label of the u-th auxiliary sample in generation t. tThe total number of auxiliary samples for cancer gene expression in generation t;

[0016] Step 3: Construct the main population and auxiliary populations;

[0017] Step 3.1: Define the current algebra as g and initialize g = 1; define the maximum algebra as g. max ;

[0018] Step 3.2: Define and randomly initialize the main population of generation g as follows: in, P represents g The nth main individual in, and express The w-th main chromosome in; if express The w-th cancer gene expression sample (x) is not selected. w ,y w ),like express Select the w-th cancer gene expression sample (x w ,y w );

[0019] Step 3.3: Define and randomly initialize the auxiliary population of the g-th generation as follows: in, Q represents g The nth auxiliary individual in, and express The u-th auxiliary chromosome in; if express The auxiliary sample for the expression of the uth cancer gene was not selected. like express Select the uth cancer gene expression auxiliary sample

[0020] Step 4: If g is not a multiple of k and r, execute steps 4.1, 4.2 and 4.3 to calculate the optimization target value of the two populations in the g-th generation, which will be used to evolve the two populations in the g+1-th generation; otherwise, execute step 5.

[0021] Step 4.1: Calculate the optimization target value for the individual;

[0022] Step 4.1.1, The corresponding cancer gene expression sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression classification model of the g-th generation. And calculate In {(x w ,yw Classification accuracy on |w=1,2,…,W} and compression ratio Thus obtain Optimization target value

[0023] Step 4.1.2, The corresponding cancer gene expression auxiliary sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression auxiliary classification model of the g-th generation. And calculate exist Classification accuracy and compression ratio Thus obtain Optimization target value

[0024] Step 4.2: Evolution of the g-th generation main population P g The main population P of generation g+1 is obtained. g+1 And record the main population at the frontier of generation g+1. in, Let f represent the f-th leading individual of the frontier in the (g+1)th generation, and F represent the size of the leading population of the frontier.

[0025] Step 4.3: Evolve the g-th generation auxiliary population Q g The auxiliary population Q of generation g+1 is obtained. g+1 And record the auxiliary population at the frontier of generation g+1. in, Let E represent the e-th frontal auxiliary individual in the (g+1)-th generation, where E represents the size of the frontal auxiliary population.

[0026] Step 5: If g is a multiple of k and g is not a multiple of r, execute the main population P. g and auxiliary population Q g The knowledge transfer between them yields the g+1th generation dominant population P. g+1 And record the main population at the frontier of generation g+1.

[0027] If g is a multiple of r, then the (g+1)th generation dominant population evolves to obtain the (g+1)th generation dominant population P. g+1 And record the main population at the frontier of generation g+1.

[0028] Step 6, when g > g max At that time, obtain the g-th... max dominant population P gmax The individual that maximizes the sum of the objective values ​​is selected as the g-th individual. max The best individual like the w-th main chromosome in select the w-th cancer gene expression sample (x w , y w ), otherwise, do not select the w-th cancer gene expression sample (x w , y w ), thereby constructing an optimal subset of cancer gene expression samples;

[0029] Step 7: Input the optimal cancer gene expression sample subset into a feature selection algorithm for processing, screen out the final cancer gene expression combination, which is the screening result of cancer markers.

[0030] The cancer marker screening method based on multi-task sample selection described in the present invention is characterized in that said step 4.2 comprises the following steps:

[0031] Step 4.2.1: When g=1, from P g randomly select the a-th main individual and the b-th main individual and

[0032] construct a g-th generation crossover probability vector wherein, represents the w-th crossover random number of the g-th generation;

[0033] define the crossover probability as pc, if randc w g < pc, then and are exchanged; otherwise, no exchange is performed;

[0034] construct the a-th mutation probability vector of the g-th generation wherein, randm_a w g represents the a-th main individual of the g-th generation 's w-th mutation random number;

[0035] define the mutation probability as pm, if randm_a w g < pm, flip 's w-th main chromosome otherwise, do not flip;

[0036] construct the b-th mutation probability vector of the g-th generation wherein, randm_b w g represents the b-th main individual of the g-th generation 's w-th random number;

[0037] if random_b w g <pm, flip the w-th main chromosome of otherwise, do not flip;

[0038] add to the g-th generation updated main population P' g ;

[0039] Step 4.2.2, Execute the process according to step 4.2.1 times, so as to obtain the g-th generation updated main population P' g , and after merging it with P g , through non-dominated sorting and elite selection strategy, select the top N individuals ranked from the merged population to form the g+1-th generation main population P g+1 , and record the g+1-th generation front main population wherein, represents the f-th front main individual of the g+1-th generation, and F represents the size of the front main population;

[0040] Step 4.2.3, when g≠1, randomly select the f-th front main individual from the g-th generation front main population PF g as 's guiding individual;

[0041] Construct the learning probability vector of the g-th generation wherein, randl w g represents the w-th random number of the g-th generation;

[0042] Define the learning probability as pl, if randl w g <pl, then the w-th main chromosome in is assigned to otherwise, no assignment is performed;

[0043] add to the g-th generation updated main population P' g ;

[0044] Step 4.2.4, after merging P g and P' g , through non-dominated sorting and elite selection strategy, select the top N main individuals ranked from the merged population to form the g+1-th generation main population P g+1 , and record the g+1-th generation front main population

[0045] ​Step 4.3 includes the following steps:

[0046] Step 4.3.1, from Q g Randomly select the a-th auxiliary individual The b-th auxiliary individual and

[0047] Constructing the crossover probability vector of the g-th generation in, This represents the u-th crossover random number in the g-th generation;

[0048] Define the crossover probability pc′, if Then and Exchange; otherwise, do not exchange.

[0049] Construct the mutation probability vector of the a-th generation in the g-th generation in, This represents the a-th auxiliary individual in the g-th generation. The u-th mutated random number;

[0050] Define the mutation probability as pm′, if Then flip The uth auxiliary chromosome Otherwise, do not flip;

[0051] Construct the mutation probability vector of the b-th generation g. in, This represents the b-th auxiliary individual in the g-th generation. The u-th mutated random number;

[0052] like Then flip The uth auxiliary chromosome Otherwise, do not flip;

[0053] Will Join the g-generation update auxiliary population Q′ g middle;

[0054] Step 4.3.2: Follow the procedure in step 4.3.1. This yields the g-th generation update auxiliary population Q′. g and with Q g After merging, the top N individuals are selected from the merged population using a non-dominated ranking and elite selection strategy to form the (g+1)th generation auxiliary population Q. g+1 And record the auxiliary population at the frontier of generation g+1. in, Let E represent the e-th frontal auxiliary individual in the g+1-th generation, where E represents the size of the frontal auxiliary population.

[0055] In step 5, if g is a multiple of k and g is not a multiple of r, then steps 5.1 and 5.2 are executed.

[0056] If g is a multiple of r, then execute steps 5.2 and 5.3:

[0057] Step 5.1: Calculate the dominant population P g Each main individual The sum of the optimization objective values and P g The principal individual corresponding to the maximum sum of the optimization objective values ​​is selected as the best principal individual in the g-th generation.

[0058] merge and Q g Then, using non-dominated ranking and elite selection strategies, the top N individuals are selected from the merged population to form the (g+1)th generation auxiliary population Q. g+1 And record the auxiliary population at the frontier of generation g+1.

[0059] according to and S t Update the flag vector of generation t+1 in, This represents the w-th flag in the flag vector of the (t+1)-th generation;

[0060] like Retain the w-th cancer gene expression sample (x) w ,y w Otherwise, delete the w-th cancer gene expression sample (x). w ,y w This allows for the construction of an auxiliary dataset for cancer gene expression in generation t+1. in, This represents the auxiliary sample for the expression of the u-th cancer gene in the (t+1)-th generation; Let represent the set of attribute features of the u-th auxiliary sample in the (t+1)-th generation. U represents the class label of the u-th auxiliary sample in the (t+1)-th generation; t+1 The total number of auxiliary samples for cancer gene expression;

[0061] Step 5.2, Merge QF g P g Furthermore, through non-dominant ranking and elite selection strategies, the top N dominant individuals are selected from the merged population to form the (g+1)th generation dominant population P. g+1 And record the main population at the frontier of generation g+1.

[0062] Step 5.3, according to PF g and S t Update the flag vector of generation t+1

[0063] like Retain the w-th cancer gene expression sample (x) w ,y w Otherwise, delete the w-th cancer gene expression sample (x). w ,y w This allows for the construction of an auxiliary dataset for cancer gene expression in generation t+1.

[0064] use Randomly construct the auxiliary population Q of generation g+1. g+1 .

[0065] The present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the cancer marker screening method, and the processor is configured to execute the program stored in the memory.

[0066] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the steps of the cancer marker screening method.

[0067] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0068] 1. The method proposed in this invention is based on multi-objective optimization, which can significantly improve the efficiency of sample selection. Two objective functions are constructed by considering the classification accuracy of the sample subset and the compression rate of the sample dataset, transforming the problem into an intuitive mathematical model. The sample subset that performs well on both objectives is then selected for screening cancer markers.

[0069] 2. This invention designs a multi-task-based sample selection method, constructing a simplified auxiliary task dataset based on the importance of cancer gene expression sample data. Two task populations are constructed and evolve independently, with knowledge transfer between them occurring over a fixed number of generations. The main task population can perform a fine-grained search across the entire cancer gene expression sample dataset, while the auxiliary task population accelerates the search efficiency and effectiveness of the main task population. This allows for the selection of a subset of samples that better represents the entire cancer gene expression sample dataset.

[0070] 3. This invention proposes a method to optimize the screening effect and efficiency of cancer markers from the perspective of improving the quality of sample datasets. Unlike existing studies, this method improves the quality of sample datasets by removing noise and erroneous samples and filtering out a large number of redundant samples. This significantly reduces the screening time while improving the screening effect of cancer markers. Attached Figure Description

[0071] Figure 1 This is a flowchart of the method of the present invention;

[0072] Figure 2 These are the cancer markers ultimately selected in this invention. Detailed Implementation

[0073] In this embodiment, a cancer marker screening method based on multi-task sample selection is described, such as... Figure 1 As shown, it includes the following steps:

[0074] Step 1: Obtain the cancer gene expression dataset {(x w ,y w )|w=1,2,…,W}, where, (x w ,y w ) represents the w-th cancer gene expression sample, x w Let w be the set of attribute features of the w-th sample, and Let y represent the d-th attribute feature of the w-th sample, where d∈[1,D], and D is the total number of attribute features of the cancer gene expression samples. w Let y represent the class label of the w-th sample. w w∈[1,C], where C represents the total number of categories; w∈[1,W], where W is the total number of cancer gene expression samples;

[0075] In this embodiment, the cancer gene expression dataset uses the gene expression cancerRNA-Seq dataset as the standard dataset. In the gene expression cancerRNA-Seq dataset, there are a total of 801 data points, corresponding to 801 independent biological samples of cancer gene expression data collected.

[0076] Step 2: Construct the auxiliary task dataset;

[0077] Step 2.1, from {(x w ,y w Obtain all samples of the c-th class from the set |w = 1, 2, ..., W}, and calculate the average of the attribute features of all samples of the c-th class, which is used as the center vector of the c-th class. c; Calculate the expression of the w-th cancer gene sample and the center c After calculating the Euclidean distance and normalizing it, the importance imp of the w-th sample is obtained. w Thus, the importance of all samples is obtained: imp = {imp} w |w=1,2,…,W};

[0078] Step 2.2: Define the algebra of the auxiliary task dataset as t, and initialize t = 1;

[0079] Step 2.2: Define the algebra of the auxiliary task dataset as t, and initialize t = 1;

[0080] Define the t-th generation flag vector in, This represents the w-th flag in the flag vector of generation t;

[0081] Construct a random probability vector pos = {pos w |w=1,2,…,W};where, pos w This represents the w-th random number;

[0082] If pos w <imp w Then let This indicates that the w-th cancer gene expression sample (x) is retained. w ,y w Otherwise, let This indicates the deletion of the w-th cancer gene expression sample (x). w ,y w This allows for the construction of a generation t auxiliary dataset for cancer gene expression. in, This represents the auxiliary sample for the expression of the u-th cancer gene in generation t; Let represent the set of attribute features of the u-th auxiliary sample in the t-th generation. U represents the class label of the u-th auxiliary sample in generation t. t The total number of auxiliary samples for cancer gene expression in generation t;

[0083] Step 3: Construct the main population and auxiliary populations;

[0084] Step 3.1: Define the current algebra as g and initialize g = 1; define the maximum algebra as g. max ;

[0085] Step 3.2: Define and randomly initialize the main population of generation g as follows: in, P represents g The nth main individual in, and express The w-th main chromosome in; if express The w-th cancer gene expression sample (x) is not selected. w ,y w ),like express Select the w-th cancer gene expression sample (x w ,y w );

[0086] Step 3.3: Define and randomly initialize the auxiliary population of the g-th generation as follows: in, Q represents g The nth auxiliary individual in, and express The u-th auxiliary chromosome in; if express The auxiliary sample for the expression of the uth cancer gene was not selected. like express Select the uth cancer gene expression auxiliary sample

[0087] Step 4: Assign 5 to k and 20 to r. If g is not a multiple of k and r, execute steps 4.1, 4.2 and 4.3 to calculate the optimization target values ​​of the two populations in the g-th generation, which will be used to evolve the two populations in the g+1-th generation.

[0088] Step 4.1: Calculate the optimization target value for the individual;

[0089] Step 4.1.1, The corresponding cancer gene expression sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression classification model of the g-th generation. And calculate In {(x w ,y w Classification accuracy on |w=1,2,…,W} and compression ratio Thus obtain Optimization target value

[0090] Step 4.1.2, The corresponding cancer gene expression auxiliary sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression auxiliary classification model of the g-th generation. And calculate exist Classification accuracy and compression ratio to obtain the optimized target value

[0091] Step 4.2: Main population P of the g-th generation g ;

[0092] Step 4.2.1: When g=1, from P g randomly select the a-th main individual and the b-th main individual and

[0093] construct the g-th generation crossover probability vector wherein, randc w g represents the w-th crossover random number of the g-th generation;

[0094] define the crossover probability as pc, assign 0.9 to pc, if randc w g < pc, then swap and ; otherwise, do not swap;

[0095] construct the a-th mutation probability vector of the g-th generation wherein, randm_a w g represents the a-th main individual of the g-th generation 's w-th mutation random number;

[0096] define the mutation probability as pm, assign 0.02 to pm, if then flip the w-th main chromosome of otherwise, do not flip;

[0097] construct the b-th mutation probability vector of the g-th generation wherein, randm_b w g represents the b-th main individual of the g-th generation 's w-th random number;

[0098] if randm_b w g < pm, then flip the w-th main chromosome of otherwise, do not flip;

[0099] add into the updated main population P' of the g-th generation g ;

[0100] Step 4.2.2: Execute the process according to step 4.2.1 times, so as to obtain the g-th generation updated main population P′ g , and after merging with P g , through non-dominated sorting and elitist selection strategy, select the top N individuals with the highest ranking from the merged population to form the (g+1)-th generation main population P g+1 , and record the (g+1)-th generation front main population wherein, represents the f-th front main individual of the (g+1)-th generation, and F represents the size of the front main population;

[0101] Step 4.2.3: When g≠1, select from the g-th generation front main population PF g the f-th front main individual randomly as the guiding individual of;

[0102] construct the learning probability vector of the g-th generation wherein, randl w g represents the w-th random number of the g-th generation;

[0103] define the learning probability as pl, assign 0.7 to pl, if randl w g < pl, then the w-th main chromosome in is assigned to otherwise, no assignment is performed;

[0104] add to the g-th generation updated main population P′ g ;

[0105] Step 4.2.4: After merging P g and P′ g , through non-dominated sorting and elitist selection strategy, select the top N main individuals with the highest ranking from the merged population to form the (g+1)-th generation main population P g+1 , and record the (g+1)-th generation front main population

[0106] Step 4.3: Evolve the g-th generation auxiliary population Q g ;

[0107] Step 4.3.1: Select from Q g the a-th auxiliary individual randomly and the b-th auxiliary individual and

[0108] construct the crossover probability vector of the g-th generation in, This represents the u-th crossover random number in the g-th generation;

[0109] Define the crossover probability pc′, and assign 0.9 to pc′. If Then and Exchange; otherwise, do not exchange.

[0110] Construct the mutation probability vector of the a-th generation in the g-th generation in, This represents the a-th auxiliary individual in the g-th generation. The u-th mutated random number;

[0111] Define the mutation probability as pm′, and assign 0.02 to pm′. If Then flip The uth auxiliary chromosome Otherwise, do not flip;

[0112] Construct the mutation probability vector of the b-th generation g. in, This represents the b-th auxiliary individual in the g-th generation. The u-th mutated random number;

[0113] like Then flip The uth auxiliary chromosome Otherwise, do not flip;

[0114] Will Join the g-generation update auxiliary population Q′ g middle;

[0115] Step 4.3.2: Follow the procedure in step 4.3.1. This yields the g-th generation update auxiliary population Q′. g and with Q g After merging, the top N individuals are selected from the merged population using a non-dominated ranking and elite selection strategy to form the (g+1)th generation auxiliary population Q. g+1 And record the auxiliary population at the frontier of generation g+1. in, Let E represent the e-th frontal auxiliary individual in the (g+1)-th generation, where E represents the size of the frontal auxiliary population.

[0116] Step 5: If g is a multiple of k and g is not a multiple of r, execute steps 5.1 and 5.2 for the main population P. g and auxiliary population Q g Knowledge transfer between them;

[0117] If g is a multiple of r, execute step 5.2 to evolve the main population of generation g+1, and execute step 5.3 to construct the auxiliary population of generation t+1.

[0118] Step 5.1: Calculate the dominant population P g Each main individual The sum of the optimization objective values and P g The principal individual corresponding to the maximum sum of the optimization objective values ​​is selected as the best principal individual in the g-th generation.

[0119] merge and Q g Then, using non-dominated ranking and elite selection strategies, the top N individuals are selected from the merged population to form the (g+1)th generation auxiliary population Q. g+1 And record the auxiliary population at the frontier of generation g+1.

[0120] according to and S t Update the flag vector of generation t+1 in, This represents the w-th flag in the flag vector of the (t+1)-th generation;

[0121] like Retain the w-th cancer gene expression sample; otherwise, delete the w-th cancer gene expression sample, thus constructing the (t+1)-th generation cancer gene expression auxiliary dataset. in, This represents the auxiliary sample for the expression of the u-th cancer gene in the (t+1)-th generation; Let represent the set of attribute features of the u-th auxiliary sample in the (t+1)-th generation. U represents the class label of the u-th auxiliary sample in the (t+1)-th generation; t+1 The total number of auxiliary samples for cancer gene expression;

[0122] Step 5.2, Merge QF g P g By employing non-dominant ranking and elite selection strategies, the top N dominant individuals are selected from the merged population to form the (g+1)th generation dominant population P. g+1 And record the main population at the frontier of generation g+1.

[0123] Step 5.3, according to PF g and S t Update the flag vector of generation t+1

[0124] like Retain the w-th cancer gene expression sample; otherwise, delete the w-th cancer gene expression sample, thus constructing the (t+1)-th generation cancer gene expression auxiliary dataset.

[0125] use Randomly construct the auxiliary population Q of generation g+1. g+1 ;

[0126] Step 6, when g > g max At that time, obtain the g-th... max dominant population P gmax The individual that maximizes the sum of the objective values ​​is selected as the g-th individual. max The best individual like The w-th main chromosome Select the w-th cancer gene expression sample (x w ,y w Otherwise, do not select the w-th cancer gene expression sample (x). w ,y w This allows for the construction of an optimal subset of cancer gene expression samples.

[0127] As shown in Table 1, the classification accuracy and compression rate of the optimal cancer gene expression sample subset are calculated. The comparison with the original dataset shows that the optimal cancer gene expression sample subset removes redundant and erroneous samples from the original sample set.

[0128] Table 1

[0129]

[0130] Step 7: Input the optimal subset of cancer gene expression samples into the feature selection algorithm for processing, and select the final combination of cancer gene expression, i.e. the screening result of cancer markers.

[0131] like Figure 2 As shown, the final binary chromosome sequence obtained by feature selection is (1,0,1,…,0,0,1), which corresponds to the set of gene expressions at positions 1, 3,…,L in the cancer gene expression dataset. This set represents potential cancer markers. Table 2 shows a comparison between the method of this invention and traditional feature selection methods on the gene expression cancer RNA-Seq dataset.

[0132] Table 2

[0133]

[0134] As shown in Table 2, compared with traditional feature selection methods, the cancer markers screened using the cancer gene expression dataset in this invention have higher classification accuracy across the entire dataset. This means that the gene expression combination corresponding to the cancer marker has a closer relationship with cancer, i.e., the screening effect is better. Furthermore, the screening time of this invention is shorter than that of traditional feature selection methods. Currently, with the development of high-throughput sequencing technology, cancer gene expression data is constantly increasing, and this invention's method can effectively improve the effectiveness and efficiency of screening cancer markers using cancer gene expression data.

Claims

1. A cancer marker screening method based on multi-task sample selection, characterized in that, Includes the following steps: Step 1: Obtain cancer gene expression dataset ,in, x represents the w-th cancer gene expression sample. w Let w be the set of attribute features of the w-th sample, and , This represents the d-th attribute feature of the w-th sample. D represents the total number of attribute features of the cancer gene expression samples, and y w This represents the class label of the w-th sample. C represents the total number of categories; W represents the total number of cancer gene expression samples; Step 2: Construct the auxiliary task dataset; Step 2.1, from We obtain all samples of the c-th category and calculate the average of the attribute features in all samples of the c-th category, which is used as the center vector of the c-th category. c ; Calculate the expression of the w-th cancer gene sample and the center c After calculating the Euclidean distance and normalizing it, the importance imp of the w-th sample is obtained. w This allows us to determine the importance of all samples. ; Step 2.2: Define the algebra of the auxiliary task dataset as t, and initialize t=1; Define the t-th generation flag vector ,in, This represents the w-th flag in the flag vector of generation t; Constructing random probability vectors ; where pos w This represents the w-th random number; If pos w <imp w Then let This indicates that the w-th cancer gene expression sample is retained. Otherwise, let This indicates the deletion of the w-th cancer gene expression sample. Thus, a generation t cancer gene expression auxiliary dataset is constructed. ,in, This represents the auxiliary sample for the expression of the u-th cancer gene in generation t; Let represent the set of attribute features of the u-th auxiliary sample in the t-th generation. This represents the class label of the u-th auxiliary sample in generation t. The total number of auxiliary samples for cancer gene expression in generation t; Step 3: Construct the main population and auxiliary populations; Step 3.1: Define the current algebra as g and initialize g=1; define the maximum algebra as g. max ; Step 3.2: Define and randomly initialize the main population of generation g as follows: ,in, express The nth main individual in, and , express The w-th main chromosome in; if =0 means The wth cancer gene expression sample was not selected. ,like =1 indicates Select the wth cancer gene expression sample ; Step 3.3: Define and randomly initialize the auxiliary population of the g-th generation as follows: ,in, express The nth auxiliary individual in, and , express The u-th auxiliary chromosome in; if =0 means The auxiliary sample for the expression of the uth cancer gene was not selected. ,like =1 indicates Select the uth cancer gene expression auxiliary sample ; Step 4, if If the value is not a multiple of k and r, execute steps 4.1, 4.2, and 4.3 to calculate the optimization target value for the two populations in the g-th generation, which will be used to evolve the two populations in the g+1-th generation; otherwise, execute step 5. Step 4.1: Calculate the optimization target value for the individual; Step 4.1.1, The corresponding cancer gene expression sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression classification model of the g-th generation. and calculate exist Classification accuracy and compression ratio Thus obtain Optimization target value ; Step 4.1.2, The corresponding cancer gene expression auxiliary sample set is input into the SVM classifier for training, resulting in the nth cancer gene expression auxiliary classification model of the g-th generation. and calculate exist Classification accuracy and compression ratio Thus obtain Optimization target value ; Step 4.2: Evolution of the g-th generation main population The g+1 generation main population was obtained. And record the main population at the frontier of generation g+1. ;in, Let f represent the f-th leading individual of the frontier in the (g+1)th generation, and F represent the size of the leading population of the frontier. Step 4.3: Evolve the g-th generation auxiliary population The g+1 generation auxiliary population was obtained. And record the auxiliary population at the frontier of generation g+1. ;in, Let E represent the e-th frontal auxiliary individual in the (g+1)-th generation, where E represents the size of the frontal auxiliary population. Step 5, if Multiples of k and Not for When the number is a multiple of the target population, execute the main population control. and auxiliary population Knowledge transfer between generations yields the g+1th generation of the main population. And record the main population at the frontier of generation g+1. ; like for When the number of generations is a multiple of , the (g+1)th generation of the dominant population evolves to obtain the (g+1)th generation of the dominant population. And record the main population at the frontier of generation g+1. ; Step 6, when g > g max At that time, obtain the g-th... max Substitute population The individual that maximizes the sum of the objective values ​​is selected as the g-th individual. max The best individual ,like The w-th main chromosome =1, select the w-th cancer gene expression sample. Otherwise, do not select the w-th cancer gene expression sample. This allows for the construction of an optimal subset of cancer gene expression samples. Step 7: Input the optimal subset of cancer gene expression samples into the feature selection algorithm for processing, and select the final combination of cancer gene expression, i.e. the screening result of cancer markers.

2. The cancer marker screening method based on multi-task sample selection according to claim 1, characterized in that, Step 4.2 includes the following steps: Step 4.2.1: When g=1, from Randomly select the a-th master individual and the b-th main individual ,and ; Constructing the crossover probability vector of the g-th generation Among them, randc w g This represents the w-th crossover random number in the g-th generation; Define the crossover probability as pc. If randc w g When < pc, then and Exchange; otherwise, do not exchange. Construct the mutation probability vector of the a-th generation in the g-th generation , where randm_a w g This represents the a-th principal individual in the g-th generation. The w-th mutated random number; Define the mutation probability as pm, if randm_a w g If it's less than pm, then flip it. The wth main chromosome Otherwise, do not flip; Construct the mutation probability vector of the b-th generation g. Among them, randm_b w g This represents the b-th principal individual in the g-th generation. The w-th random number; If randm_b w g If it's less than pm, then flip it. The wth main chromosome Otherwise, do not flip; Will , Join the gth generation update main population middle; Step 4.2.2: Follow the procedure in step 4.2.

1. This leads to the g-th generation update of the primary population. and with After merging, the top N individuals are selected from the merged population using a non-dominated ranking and elite selection strategy to form the main population of generation g+1. And record the main population at the frontier of generation g+1. ;in, Let f represent the f-th leading individual of the frontier in the (g+1)th generation, and F represent the size of the leading population of the frontier. Step 4.2.3, when At that time, from the g-th generation frontal main population Randomly select the f-th leading edge individual As The guiding individual; Constructing the learning probability vector for the g-th generation , where randl w g This represents the w-th random number in the g-th generation; Define the learning probability as p1, if randl w g When < pl, then The w-th main chromosome Assign to Otherwise, no value will be assigned; Will Join the gth generation update main population middle; Step 4.2.4, and After merging, the top N dominant individuals from the merged population are selected using a non-dominant ranking and elite selection strategy to form the (g+1)th generation dominant population. And record the main population at the frontier of generation g+1. .

3. The cancer marker screening method based on multi-task sample selection according to claim 2, characterized in that, Step 4.3 includes the following steps: Step 4.3.1, from Randomly select the a-th auxiliary individual The bth auxiliary individual ,and ; Constructing the crossover probability vector of the g-th generation ,in, This represents the u-th crossover random number in the g-th generation; Define crossover probability ,like Then and Exchange; otherwise, do not exchange. Construct the mutation probability vector of the a-th generation in the g-th generation ,in, This represents the a-th auxiliary individual in the g-th generation. The u-th mutated random number; Define the mutation probability as ,like Then flip The uth auxiliary chromosome Otherwise, do not flip; Construct the mutation probability vector of the b-th generation g. ,in, This represents the b-th auxiliary individual in the g-th generation. The u-th mutated random number; like Then flip The uth auxiliary chromosome Otherwise, do not flip; Will , Join the gth generation update auxiliary population middle; Step 4.3.2: Follow the procedure in step 4.3.

1. This process yields the g-th generation updated auxiliary population. and with After merging, the top N individuals are selected from the merged population using a non-dominated ranking and elite selection strategy to form the (g+1)th generation auxiliary population. And record the auxiliary population at the frontier of generation g+1. ;in, Let E represent the e-th frontal auxiliary individual in the g+1-th generation, where E represents the size of the frontal auxiliary population.

4. The cancer marker screening method based on multi-task sample selection according to claim 3, characterized in that, In step 5, if Multiples of k and Not for If the value is a multiple of the given value, then proceed with steps 5.1 and 5.2; if... for If the value is a multiple of the given value, then proceed with steps 5.2 and 5.3: Step 5.1: Calculate the dominant population Each main individual The sum of the optimization objective values and will The principal individual corresponding to the maximum sum of the optimization objective values ​​is selected as the best principal individual in the g-th generation. ; merge and Then, using non-dominated ranking and elite selection strategies, the top N individuals are selected from the merged population to form the g+1th generation auxiliary population. And record the auxiliary population at the frontier of generation g+1. ; according to and Update the flag vector of generation t+1 ,in, This represents the w-th flag in the flag vector of the (t+1)-th generation; like Preserve the wth cancer gene expression sample Otherwise, delete the w-th cancer gene expression sample. Thus, a cancer gene expression auxiliary dataset of generation t+1 is constructed. ; in, This represents the auxiliary sample for the expression of the u-th cancer gene in the (t+1)-th generation; Let represent the set of attribute features of the u-th auxiliary sample in the (t+1)-th generation. This represents the class label of the u-th auxiliary sample in the (t+1)-th generation; The total number of auxiliary samples for cancer gene expression; Step 5.2, Merge , Furthermore, through non-dominant ranking and elite selection strategies, the top N dominant individuals are selected from the merged population to form the (g+1)th generation dominant population. And record the main population at the frontier of generation g+1. ; Step 5.3, according to and Update the flag vector of generation t+1 ; like Preserve the wth cancer gene expression sample Otherwise, delete the w-th cancer gene expression sample. Thus, a cancer gene expression auxiliary dataset of generation t+1 is constructed. ; use Randomly construct the auxiliary population of generation g+1. .

5. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing any of the cancer marker screening methods of claims 1-4, the processor being configured to execute the program stored in the memory.

6. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program, when run by a processor, performs the steps of the cancer marker screening method according to any one of claims 1-4.

Images