A microwell screening system based on consistency constraints

Abstract

The application discloses a microwell screening system based on consistency constraint, comprising: a data acquisition module for acquiring the viability value data of the cultured sample in the well of the multi-well plate; a classification processing module for classifying the viability value data to obtain multiple data groups; a statistical index acquisition module for calculating the coefficient of variation of the multiple data groups respectively, and taking the coefficient of variation as a statistical dispersion index; a judgment standard module for judging whether the data group reaches the standard of consistency constraint according to the comparison result of the statistical dispersion index and the preset consistency threshold, and the number of elements in the data group. The application can realize fast and effective consistency evaluation and improvement, improve the robustness of the experiment, control the uniformity of the experimental sample, save time and effort, exclude the interference of subjective factors, exclude the objective error of the culture system, help to utilize the high-throughput well plate in layers, and help to improve the comprehensive efficiency of the cultured sample in the subsequent culture and experiment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biological medicine, and particularly relates to a microwell screening system based on consistency constraint. BACKGROUND

[0002] In biological and medical analysis experiments, multiple samples are often cultured and analyzed simultaneously using multi-well plates. With the advancement of biological manufacturing technology, the throughput of multi-well plates is gradually increasing. Culturing of various biological microspheres such as cell lines and organoids is sometimes based on 96-well plates, 384-well plates, or even 1536-well plates.

[0003] There is a certain heterogeneity in the growth of microspheres such as tumors, and factors such as cultivation system deviation and human operation can cause different degrees of inter-well differences in the cultured samples. Therefore, in order to improve the robustness of the experiment, it is necessary to perform certain quality control on the uniformity of the samples, so as to screen out sample microwells that meet the consistency requirements. Existing microwell screening methods mostly regard all the same test objects on the entire multi-well plate as a group and perform consistency evaluation, and often rely on manual identification of abnormal wells and evaluation of the uniformity of microwells. However, this method is relatively time-consuming and labor-intensive, may be disturbed by human subjective factors, cannot accurately test the consistency degree of the sample, and lacks utilization of the potential natural group stratification of the sample to be tested. SUMMARY

[0004] The present application relates to the technical field of biological medicine, and particularly relates to a microwell screening system based on consistency constraint.

[0005] The technical solution adopted by the present application is as follows:

[0006] A microwell screening system based on consistency constraint comprises:

[0007] A data acquisition module is configured to acquire viability value data of a cultured sample in a well of a multi-well plate.

[0008] A classification processing module is configured to perform classification processing on the viability value data to obtain a plurality of data groups.

[0009] A statistical index acquisition module is configured to calculate a coefficient of variation of each of the plurality of data groups, and take the coefficient of variation as a statistical dispersion index.

[0010] A judgment standard module is configured to judge whether the data group meets a consistency constraint standard according to a comparison result of the statistical dispersion index and a preset consistency threshold, and a number of elements in the data group.

[0011] As preferred, the function flow of the data acquisition module is as follows:

[0012] The ATP method or microscopic image is used to extract the viability value data of the sample cultured in the well of the multi-well plate.

[0013] As preferred, the function flow of the classification processing module is as follows:

[0014] The viability value data is classified to obtain m groups of classification groups, wherein m≥1;

[0015] When m=1, the classification groups are taken as the data groups.

[0016] When m≥2, the k-means clustering method is used to cluster the viability value data into m groups, to obtain m groups of clustered classification groups, and the m groups of clustered classification groups are taken as the data groups.

[0017] As preferred, the function flow of the judgment standard module is as follows:

[0018] The consistency constraints of the plurality of data groups are judged one by one:

[0019] When the statistical dispersion index of a data group is less than the preset consistency threshold,

[0020] If the number of elements in the data group is greater than or equal to the preset number, the data group meets the consistency constraint standard.

[0021] Otherwise, the data group does not meet the consistency constraint standard.

[0022] As preferred, when the statistical dispersion index of a data group is not less than the preset consistency threshold, the system further comprises the following function flow:

[0023] Step S1: According to the data distribution of the data group, a corresponding outlier detection method is selected;

[0024] Step S2: A detection threshold is set, and outliers are detected according to the outlier detection method, the outliers are excluded from the data group, and a new data group is obtained;

[0025] Step S3: The coefficient of variation of the new data group is calculated, and the coefficient of variation of the new data group is taken as a new statistical dispersion index;

[0026] Step S4: when the new statistical dispersion index is less than the preset consistency threshold and the number of elements in the new data set is greater than or equal to a preset number, the data set reaches the standard of consistency constraint, and consistency judgment of the next data set is entered; if the number of elements in the new data set is less than the preset number, the new data set does not reach the standard of consistency constraint, and consistency judgment of the next data set is entered.

[0027] When the new statistical dispersion index is not less than the preset consistency threshold, step S5 is entered.

[0028] Step S5: the detection threshold is adjusted by a preset step value to obtain a new detection threshold, and steps S3-S4 are repeated until the new detection threshold exceeds a critical value, the detection of outliers is stopped, and it is judged that the current data set cannot reach the standard of consistency constraint.

[0029] As a preferred, the step S1 comprises:

[0030] When the number of vitality value data in the data set is less than or equal to a preset number or the vitality value data in the data set is not normally distributed, the outlier detection method selects the mean method or the median method.

[0031] When the vitality value data in the data set is normally distributed, the outlier detection method selects the Grubbs test method or the Studentized deviation test method.

[0032] As a preferred, the mean method is that elements greater than (1+1.5)*upper quartile or less than (1-1.5)*lower quartile in the data set are taken as outliers.

[0033] As a preferred, the median method is:

[0034] The median in the data set is converted to obtain a three-fold median absolute deviation.

[0035] Elements in the data set whose difference from the median is greater than the three-fold median absolute deviation are taken as outliers.

[0036] The beneficial effects of the present application are that the present application can realize fast and effective consistency evaluation and improvement, improve the robustness of experiments, control the uniformity of experimental samples, save time and effort, can exclude the interference of subjective factors, exclude the objective errors of the culture system, help to perform hierarchical utilization on high-throughput well plates, and is conducive to improving the comprehensive performance of the detected culture samples in subsequent culture and experiments. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1This is a functional flowchart of a micropore screening system based on consistency constraints according to the present invention.

[0038] Figure 2 This is a discrete diagram of the two groups before and after screening in Embodiment 2 of the present invention. Detailed Implementation

[0039] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0040] See Figure 1 A micropore screening system based on consistency constraints, comprising:

[0041] Data acquisition module: used to acquire the viability data of cultured samples in the wells of a multi-well plate;

[0042] The activity value data is any selected index used to characterize the target activity within the well. Preferably, the activity value data is the maximum cross-sectional area, fluorescence intensity after staining with reactive dye, ATP intensity, etc.

[0043] The functional flow of the data acquisition module is as follows:

[0044] Viability data of cultured samples in wells of multi-well plates were obtained using the ATP method or extracted from microscopic images.

[0045] Summarize the vitality values ​​of all pores X = {x1, x2, ..., x n}, where x i The vitality value data corresponds to the i-th well, and n is the total number of wells to be tested (n>3).

[0046] Classification processing module: used to classify the vitality value data to obtain multiple data groups;

[0047] The functional flow of the classification processing module is as follows:

[0048] The vitality value data is classified to obtain m classification groups, where m ≥ 1;

[0049] When m=1, the classification group is taken as the data group;

[0050] When m ≥ 2, the vitality value data is clustered into m groups using the k-means clustering method, resulting in m clustered classification groups. These m clustered classification groups are then used as the data groups. The clustered data groups are denoted as Y. j , where j = 1, 2, ..., m.

[0051] a statistical index acquisition module configured to calculate a coefficient of variation of each of the plurality of data groups as a statistical dispersion index;

[0052] Y j = {x j1 ,x j2 ,…,x jk}, where j k is the total number of samples in the Y j group (j k ≥ 2), and a coefficient of variation CV is calculated as a statistical dispersion index, specifically:

[0053] CV = σ / μ × 100

[0054] where σ is the standard deviation and μ is the mean

[0055] a judgment standard module configured to determine whether the data group meets the consistency constraint according to a comparison result of the statistical dispersion index and a preset consistency threshold, and a number of elements in the data group.

[0056] The function flow of the judgment standard module is as follows:

[0057] The consistency constraint of each of the plurality of data groups is determined one by one:

[0058] When the statistical dispersion index of the data group is less than the preset consistency threshold,

[0059] If the number of elements in the data group is greater than or equal to a preset number, the data group meets the consistency constraint.

[0060] Otherwise, the data group does not meet the consistency constraint.

[0061] If the coefficient of variation CV is less than the preset consistency threshold δ, it is considered that the consistency constraint is met and can be applied to future repetitive experiments. Usually δ ∈ [0, 100%], and the preset consistency threshold of uniform samples is preferably δ = 15%.

[0062] When the statistical dispersion index of the data group is not less than the preset consistency threshold, the system further includes the following function flow:

[0063] Step S1: According to the data distribution of the data group, a corresponding outlier detection method is selected;

[0064] When the number of the vitality value data in the data set is less than or equal to a preset number or the vitality value data in the data set is not normally distributed, the outlier detection method selects a mean method or a median method;

[0065] The mean method is: the elements greater than (1+1.5)*upper quartile or less than (1-1.5)*lower quartile in the data set are taken as outliers.

[0066] The median method is:

[0067] The triple median absolute deviation is obtained by converting the median in the data set.

[0068] The median absolute deviation MAD is:

[0069]

[0070] Wherein, erfcinv() refers to the inverse complementary error function of the normal distribution.

[0071] The elements in the data set whose difference from the median is greater than the triple median absolute deviation are taken as outliers.

[0072] When the vitality value data in the data set is normally distributed, the outlier detection method selects Grubbs test method or Studentized deviation test method.

[0073] Step S2: setting a detection threshold, detecting outliers according to the outlier detection method, screening out the outliers from the data set to obtain a new data set;

[0074] Step S3: calculating the coefficient of variation of the new data set, taking the coefficient of variation of the new data set as a new statistical dispersion index;

[0075] Step S4: when the new statistical dispersion index is less than the preset consistency threshold: if the number of elements in the new data set is greater than or equal to a preset number, the data set reaches the standard of consistency constraint, and enters the consistency judgment of the next data set; if the number of elements in the new data set is less than the preset number, the new data set does not reach the standard of consistency constraint, and enters the consistency judgment of the next data set;

[0076] When the new statistical dispersion index is not less than the preset consistency threshold, step S5 is entered;

[0077] Step S5: Adjust the detection threshold by a preset step size to obtain a new detection threshold. Repeat steps S3 and S4 until the new detection threshold exceeds the critical value. Stop detecting outliers and determine that the current data group cannot meet the consistency constraint standard. Specifically, for the mean and median methods, the detection threshold is reduced by a preset step size to obtain a new detection threshold; for methods based on normal distribution such as gesd and student, the detection threshold is increased by a preset step size to obtain a new detection threshold.

[0078] Initialize the parameters; taking the Student bias test as an example, set the detection threshold to 0.01. For data group Y... j Outlier detection was performed, and the detected outliers were removed to obtain a new dataset Y. j ′ The coefficient of variation (CV) is calculated. If the current CV value does not meet the preset consistency threshold δ, the detection threshold is increased by a certain step size (e.g., 0.01). (The larger the detection threshold, the more outliers are found.) The process continues for group Y. j Perform outlier testing. Iterate the above process until group Y is reached. j ′ CV < δ, or Y j ′ Number of elements in the group j k ≤3, or the detection threshold is out of range (usually the detection threshold is a value greater than 0 and less than 1).

[0079] Group Y j Outlier testing was performed, and the detected outliers were removed to obtain the new group Y. j ′ Iterate the above process until its CV < δ, or Y j ′ Number of elements in the group j k ≤3.

[0080] Group the data Y1 for all iterations in the completed steps. ′ Y2 ′ ,...,Y m ′ By summarizing, if CV < δ, a set of holes Y is considered to meet the consistency criteria. j ′ This can be used for the next experiment; if the number of elements in the group is j k If the value is ≤3 and CV≥δ, then the sample size is determined to be too large and not suitable for further experimentation.

[0081] Example 1

[0082] The homogenous organoids samples cultured in the wells of the 384-well plate are extracted from the microscopic images, the organoid targets are extracted, and the fluorescence intensity values in the wells are counted to obtain the viability value data of the wells;

[0083] Each well is sequentially numbered, and the corresponding viability value data are x1, x2,..., x n , n = 384, to obtain the viability value data set X = {x1, x2,..., x n}.

[0084] The viability value data are classified to obtain 3 groups of classification groups, and the preset threshold value δ of the coefficient of variation CV reaching the consistency standard is set to 15%.

[0085] The k-means clustering method is used to cluster the viability value data into 3 groups, and 3 groups of classification groups after clustering are obtained. The coefficient of variation of the data group Y j is calculated, and the formula is as follows:

[0086] CV = σ / μ × 100

[0087] Where σ is the standard deviation of Y j , and μ is the mean of Y j . The calculation formulas are as follows: Mean The coefficient of variation of each group obtained by calculation is shown in Table 1 below. It is found that the CV values of groups Y1, Y2 and Y3 are 12%, 8% and 16.4%, respectively. Among them, 12.2% < 15%, 8.67% < 15%, so the groups Y1 and Y2 obtained by clustering meet the consistency standard and can be reserved for use. And 16.4% > 15%, the group Y3 after clustering still does not meet the consistency standard.

[0088] Table 1 Coefficient of variation

[0089] [Y1] [Y2] [Y3] Number of elements 101 153 130 CV 12.2％ 8.67％ 16.4％

[0090] The group Y3 is extracted alone, the parameter'method' for detecting outliers is set to Studentized Deviation Test method 'gesd', the initial parameter detection threshold factor 'ThresholdFactor' is set to 0.01, and the outliers are found. The detected outliers are removed from the group to obtain a new data group Y3 ′ , and the CV of which is calculated to be 15.9%, and 15.9% > 15%, which still does not meet the set consistency constraint standard. With a step of 0.01, the detection threshold ThresholdFactor is increased to 0.02, and the gesd method is used to continue to find Y3 ′outliers in the group Y3, and calculate the CV value after removing outliers, and evaluate whether it meets the consistency constraint standard. Iterate the above process until the CV value reaches 14.7% when the detection threshold ThresholdFactor reaches 0.05, 14.7% < 15%, and check that the number of elements in the current group is 98. Based on the above three conditions, the group Y3' meets the consistency standard.

[0091] Summarize Y1, Y2 and Y3, finally, the groups Y1, Y2 and the screened group Y3' meet the consistency constraint standard, and record the group number and the well number in the group for future use. These groups are more suitable for subsequent experiments, especially repetitive experiments such as growth rate evaluation, drug sensitivity response experiment, etc.

[0092] Example 2

[0093] Using the homologous organoids cultured in the wells of the 96-well plate extracted from the microscopic images, the Calcein AM fluorescent dye is used for staining, the organoid target is extracted, and the fluorescence intensity value in the well is obtained as the viability value data of the well plate;

[0094] Each well is sequentially numbered to obtain the viability value data set X = {W i |W2,W3,W4,...}, where the row number W ∈ {B,C,D,E,F,G}, and the column number i = 2,3,...,11. The total number of elements of the viability value data set X is n = 96. See Table 2 and Table 3 for the corresponding number of the well plate and the viability value data in the well.

[0095] Table 2 Corresponding number of well plate

[0096] Well plate 1 2 3 4 5 6 7 8 9 10 11 12 A B B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 C C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 D D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 E E2 E3 E4 E5 E6 E7 E8 E9 E10 E11 F F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 G G2 G3 G4 G5 G6 G7 G8 G9 G10 G11 H

[0097] Table 3 Viability value data of well plate

[0098] Well plate 1 2 3 4 5 6 7 8 9 10 11 12 A B 111 133 143 76 111 524 220 202 98 221 C 95 94 0 74 76 66 66 75 87 48 D 60 86 94 86 154 147 40 33 64 96 E 23 19 21 10 75 84 107 71 74 90 F 63 99 54 88 73 70 155 87 109 130 G 78 76 58 61 47 71 38 67 44 63 H

[0099] Classify the viability value data to obtain a classified group of 1 group, and the group to be detected is named Y, Y = X, and the preset threshold value δ of the coefficient of variation CV reaching the consistency standard is set to 20%.

[0100] It is calculated that the CV value of the data group Y is 79%. Among them, the standard deviation σ of the data group Y is about 72.31, and the mean μ is 91.42.

[0101] The outlier detection is performed, the parameter'method' for detecting outliers is set as the median method'median', the initial parameter detection threshold factor 'ThresholdFactor' is set as 3, and outliers are found. The detected outliers are removed from the group to obtain a new data group Y3', the downlink step value is set as 0.1, and the lower limit of the detection threshold is set as 1.5. The number of removed outliers, the number of remaining elements and the current CV value in each iteration are shown in Table 4 as follows.

[0102] Table 4 Number of remaining elements and current CV value

[0103]

[0104]

[0105] In the process of iterative detection and removal of outliers, the number of detected outliers gradually increases as the detection threshold is tightened, and the consistency of the remaining elements gradually increases until CV = 18.55% < 20% = δ, and the iteration is terminated.

[0106] Collecting the above results, the CV value of the iteration termination is 18.55%, the detection threshold parameter ThresholdFactor = 1.7, and the number of remaining elements is 37. Comparing the current three parameters, CV < δ, ThresholdFactor > 1.5, and the number of elements > 3, all meet the discrimination criteria, and the screened data group Y' meets the preset consistency standard and is suitable for subsequent experiments. The retention of the micropore sample after detection is shown in Table 5.

[0107] Table 5 Retention of micropore sample after detection

[0108] Well plate 1 2 3 4 5 6 7 8 9 10 11 12 A B 111 133 143 76 111 524 220 202 98 221 C 95 94 0 74 76 66 66 75 87 48 D 60 86 94 86 154 147 40 33 64 96 E 23 19 21 10 75 84 107 71 74 90 F 63 99 54 88 73 70 155 87 109 130 G 78 76 58 61 47 71 38 67 44 63 H

[0109] Collecting the data group Y' and its elements discriminated above, and comparing them with the data group Y without detection, the improvement effect of Y' on subsequent experiments is evaluated. Figure 2 For the scatter plots of the two groups before and after screening, Table 6 shows the mean, dispersion and other statistical data before and after screening. From Figure 2 As can be seen from Table 6, the data group Y' screened by the consistency constraint method excludes micropores that are too large or too small in the data set, and has a smaller dispersion degree than the original group Y, and is more stable.

[0110] Table 6 Statistical data before and after screening

[0111] Indicator Y Y' Number of elements remaining 60 37 Mean 91.42 75.41 Standard deviation 72.31 13.99 Standard mean error 9.335 2.299 Coefficient of variation 0.791 0.1855

[0112] The above merely provides the preferred embodiments of the present application, and is not used to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the principles and technical scope of the present application shall fall into the scope of the present application.

Claims

1. A micropore screening system based on consistency constraints, characterized in that: include: Data acquisition module: used to acquire the viability data of cultured samples in the wells of a multi-well plate; Classification processing module: used to classify the vitality value data to obtain multiple data groups; The functional flow of the classification processing module is as follows: classify the vitality value data to obtain m groups of classification categories, where m≥1; when m=1, the classification category is used as the data group; when m≥2, the vitality value data is clustered into m groups using the k-means clustering method to obtain the clustered m groups of classification categories, and the clustered m groups of classification categories are used as the data group. Statistical indicator acquisition module: used to calculate the coefficient of variation of multiple sets of data respectively, and use the coefficient of variation as a statistical dispersion index; Judgment Criteria Module: Used to determine whether the data group meets the consistency constraint criteria based on the comparison result between the statistical dispersion index and the preset consistency threshold, as well as the number of elements in the data group; When the statistical dispersion index of the data set is not less than the preset consistency threshold, the system further includes the following functional flow: Step S1: Select the corresponding outlier detection method based on the data distribution of the data group; Step S2: Set a detection threshold, detect outliers according to the outlier detection method, filter out the outliers from the data group, and obtain a new data group; Step S3: Calculate the coefficient of variation of the new data set, and use the coefficient of variation of the new data set as a new statistical dispersion index; Step S4: When the new statistical dispersion index is less than the preset consistency threshold: if the number of elements in the new data group is greater than or equal to the preset number, the data group meets the consistency constraint standard and proceeds to the consistency judgment of the next data group; if the number of elements in the new data group is less than the preset number, the new data group does not meet the consistency constraint standard and proceeds to the consistency judgment of the next data group; when the new statistical dispersion index is not less than the preset consistency threshold, proceed to step S5. Step S5: Adjust the detection threshold with a preset step size to obtain a new detection threshold. Repeat steps S3-S4 until the new detection threshold exceeds the critical value. Stop detecting outliers and determine that the current data group cannot meet the consistency constraint standard.

2. The micropore screening system based on consistency constraints according to claim 1, characterized in that, The functional flow of the data acquisition module is as follows: Viability data of cultured samples in wells of multi-well plates were obtained using the ATP method or extracted from microscopic images.

3. The micropore screening system based on consistency constraints according to claim 1, characterized in that, The functional flow of the judgment criteria module is as follows: The consistency constraints of the multiple data groups are determined one by one: if the statistical dispersion index of the data group is less than the preset consistency threshold, and the number of elements in the data group is greater than or equal to the preset number, then the data group meets the consistency constraint standard; otherwise, the data group does not meet the consistency constraint standard.

4. The micropore screening system based on consistency constraints according to claim 1, characterized in that, Step S1, based on the data distribution of the data group, selects the corresponding outlier detection method to find outliers, including: If the number of vitality values ​​in the data group is less than or equal to the preset number or the vitality values ​​in the data group are not normally distributed, the outlier detection method shall be the mean method or the median method. When the vitality values ​​within the data set are normally distributed, the outlier detection method should be either the Grubbs test or the Student's deviation test.

5. A micropore screening system based on consistency constraints according to claim 4, characterized in that, The mean method is as follows: elements in the data set that are greater than (1+1.5)*upper quartile or less than (1-1.5)*lower quartile are considered outliers.

6. A micropore screening system based on consistency constraints according to claim 4, characterized in that, The median method is as follows: the median in the data set is converted to obtain three times the absolute deviation of the median; elements in the data set whose difference from the median is greater than three times the absolute deviation of the median are regarded as outliers.

Images