Computer device for monitoring minimal residual disease in multiple myeloma

Abstract

The present invention discloses a computer device for monitoring minimal residual disease in multiple myeloma, and belongs to the field of medical care informatics. The technical problem to be solved by the present invention is how to achieve high specificity and / or high sensitivity monitoring of minimal residual disease in multiple myeloma. The present invention provides a computer device, which includes a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the following steps: S1, receiving TSS relative coverage data of 50 target genes of a multiple myeloma patient to be tested; S2, inputting the data into a multiple myeloma minimal residual disease monitoring model; S3, outputting the probability value of minimal residual disease in multiple myeloma through the model; and then deriving the MRD status of the multiple myeloma patient to be tested based on the probability value. The present invention is used for monitoring minimal residual disease in multiple myeloma.

Description

Technical Field

[0001] The present invention belongs to the field of medical care informatics and relates to a computer device for monitoring minimal residual disease in multiple myeloma. Background Art

[0002] Multiple myeloma (MM) is an incurable disease. The root cause of its recurrence is the trace amount of tumor cells remaining after treatment, also known as minor residual disease (MRD). The depth of disease remission after treatment is closely related to the prognosis of patients. Patients who achieve MRD negativity have longer progression-free survival (PFS) and overall survival (OS), and the efficacy of autologous stem cell transplantation is also better. Achieving sustained MRD negativity has become the goal of treatment in the current era of new drugs, and clinical trials have used MRD as a surrogate for PFS. Monitoring MRD is of great significance for evaluating the depth of disease remission, predicting disease recurrence, and guiding treatment selection.

[0003] According to the latest response criteria published by the International Myeloma Working Group, MRD can be detected by next generation flow (NGF) or next generation sequencing (NGS). The use of NGS for MRD detection requires the collection of bone marrow specimens at the time of initial diagnosis to establish the IgH immune repertoire to detect variable regions, and bone marrow specimens must also be collected for measurement during subsequent follow-up. For patients who failed to retain bone marrow samples at the time of initial diagnosis or whose baseline gene sequences (tumor cell clones) are unclear, the NGS method cannot be used for post-treatment MRD monitoring. Therefore, NGS is only suitable for about 90% of patients, and NGS has not yet formed a standardized operating platform in China, and the interpretation of results varies on different platforms. In addition, the high price and long time-consuming nature of NGS (requiring 1-2 weeks) have prevented it from being popularized in routine clinical practice. Although NGF already has standardized reagents, instruments and antibody combinations that can ensure the consistency of results among different testing centers and the universality of evaluation standards, it takes a short time (3-4 hours) and does not require the immune phenotype at the time of initial diagnosis, but its detection requires fresh bone marrow specimens. Samples that are more than 48 hours old or frozen cannot be tested for NGF. In addition, standardized NGF requires specialized and relatively expensive instruments, analysis software and standardized test kits.

[0004] At present, the multi-parameter flow cytometry (MFC) method is commonly used in domestic clinical work to monitor the MRD of MM patients. Although the sensitivity of this technology is not as good as NGF and NGS, the instrument setting required for this technology is simpler, the detection time is short (2 hours), and it has better repeatability and lower cost, making it more suitable for current clinical practice. However, due to the differences in detection instruments and analysis in various centers, it is impossible to standardize. Finally, no matter which of the above methods is used to detect MRD, a certain amount of bone marrow samples need to be obtained through repeated bone marrow puncture for detection. However, MM patients do not need to undergo frequent bone marrow punctures except for MRD detection during regular follow-up. This method of obtaining samples will cause certain trauma and pain to patients. However, due to the trauma, poor universality or poor sensitivity of existing MRD detection methods, MRD detection of MM patients cannot be standardized according to consensus guidelines, which makes it impossible for clinicians to choose treatment and judge prognosis under the guidance of MRD. Therefore, a non-invasive, widely popularizable and more sensitive MRD detection technology is urgently needed to meet the urgent clinical needs for MRD monitoring of MM patients. Summary of the invention

[0005] The technical problem to be solved by the present invention is how to monitor minimal residual disease in multiple myeloma with high specificity and / or high sensitivity.

[0006] In order to solve the above technical problems, in a first aspect, the present invention provides a computer device for monitoring minimal residual disease in multiple myeloma.

[0007] The computer device provided by the present invention comprises a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the following steps:

[0008] S1. Receive data: Receive sample data, wherein the sample data is TSS relative coverage data of 50 target genes of a multiple myeloma patient to be tested;

[0009] The 50 target genes are ADGRL2, CNN3, COL11A1, CD244, SDC1, TMEM37, MAP2, CNTN4, ENTPD3, FAM3D, FBXO40, TMEM108, WWTR1, LINC01322, NAALADL2, KCNMB2-AS1, LINC01206, GMNC, SHROOM3, SOWAHB, LRAT, ATP10B, SAMD3, MOXD 1. ANLN, FOXP2, PTPRZ1, TUSC3, LINC01605, KCNQ3, ANO3, MUC15, BARX2, SCNN1A, PTHLH, LRIG3-DT, LINC00 402, RHOV, DLL4, ITPKA, FBN1-DT, ONECUT1, C2CD4A, ISLR2, CHRNB4, COL1A1, LAMA3, PCAT19, GGT5 and SRPX2;

[0010] The calculation formula for the relative coverage of the TSS of each target gene is as follows:

[0011]

[0012] Formula I;

[0013] In the formula, i represents the position of the TSS site of the target gene on the reference genome.

[0014] represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 2, where TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp downstream of the TSS site of the target gene to 599 bp downstream of the TSS site;

[0015] S2. Input data: input the sample data into the multiple myeloma minimal residual disease monitoring model;

[0016] The multiple myeloma minimal residual disease monitoring model is constructed according to the following steps: using the TSS relative coverage data of the 50 target genes and the corresponding minimal residual disease (MRD) status of known multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects as training samples to train the multiple myeloma minimal residual disease monitoring model;

[0017] The minimal residual disease status is multiple myeloma minimal residual disease negative or multiple myeloma minimal residual disease positive;

[0018] S3. Output result: output the probability value of whether the multiple myeloma patient to be tested is minimal residual disease positive or minimal residual disease negative through the multiple myeloma minimal residual disease monitoring model; and then derive the minimal residual disease status of the multiple myeloma patient to be tested based on the probability value.

[0019] In the above, the multiple myeloma patient to be tested may be Chinese.

[0020] In the above, the age of the multiple myeloma patient to be tested may be 18-80 years old.

[0021] In the above, the multiple myeloma patient to be tested may be a patient who has entered a remission period after treatment.

[0022] In step S1 above, the coverage may be the sequencing depth of the corresponding region calculated by the Samtools tool.

[0023] The sequencing data depth in the TSS relative coverage of each target gene can be obtained by capturing the sequencing data of the corresponding target gene in the peripheral blood plasma cfDNA of the subject or patient through a probe that specifically binds to the corresponding target gene;

[0024] Among them, the target region of the probe is the 400 target points shown in Example Table 3; the probe is an RNA probe; the 400 RNA probes corresponding to the 50 target genes are single-stranded RNA encoded by the reverse complementary sequences of the 400 target points shown in Table 3.

[0025] The TSS (transcription start site) position above is defined according to the human TSS (transcription start site) position provided by the human reference genome hg38 version;

[0026] The preparation method of the multiple myeloma minimal residual disease monitoring model in step S2 above can be specifically as follows: the modeling input data is executed using the R package 'caret' to perform a machine learning algorithm that integrates the machine learning random forest algorithm and the gradient boosting machine algorithm to establish a multiple myeloma minimal residual disease monitoring model; the model training is iteratively optimized using a three-fold cross-validation method. The optimal parameters of the model are as follows: the number of trees is 1500, the maximum depth of each tree is 15, the maximum number of features considered for each split is auto, the minimum number of samples that a node can continue to split is 5, and the minimum number of samples required for a leaf node is 3.

[0027] In an embodiment of the present invention, 70 subjects whose pathological detection of multiple myeloma minimal residual disease was negative and 70 subjects whose pathological detection of multiple myeloma minimal residual disease was positive were used as training set samples for modeling.

[0028] The model in step S3 above outputs a probability value (ranging from 0 to 1) of whether each of the multiple myeloma patients to be tested is minimal residual disease positive or minimal residual disease negative.

[0029] The step S3 above of deriving the MRD status of the multiple myeloma patient to be tested based on the probability value can be specifically as follows in an embodiment of the present invention: the model threshold is 0.53. If the model-predicted probability value of the multiple myeloma patient to be tested is greater than 0.53, the multiple myeloma patient to be tested is judged to be or is a candidate for minimal residual disease-positive. If the model-predicted probability value of the multiple myeloma patient to be tested is less than or equal to 0.53, the multiple myeloma patient to be tested is judged to be or is a candidate for minimal residual disease-negative.

[0030] In a second aspect, the present invention provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps in the computer device described in the first aspect.

[0031] The computer program product mentioned above may be a software product that implements its solution mainly through a computer program.

[0032] In a third aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the steps in the computer device according to the first aspect.

[0033] In a fourth aspect, the present invention provides a device for monitoring minimal residual disease in multiple myeloma, the device comprising:

[0034] S1, data receiving module: used to receive sample data, wherein the sample data is TSS relative coverage data of 50 target genes of multiple myeloma patients to be tested;

[0035] The 50 target genes are ADGRL2, CNN3, COL11A1, CD244, SDC1, TMEM37, MAP2, CNTN4, ENTPD3, FAM3D, FBXO40, TMEM108, WWTR1, LINC01322, NAALADL2, KCNMB2-AS1, LINC01206, GMNC, SHROOM3, SOWAHB, LRAT, ATP10B, SAMD3, MOXD 1. ANLN, FOXP2, PTPRZ1, TUSC3, LINC01605, KCNQ3, ANO3, MUC15, BARX2, SCNN1A, PTHLH, LRIG3-DT, LINC00 402, RHOV, DLL4, ITPKA, FBN1-DT, ONECUT1, C2CD4A, ISLR2, CHRNB4, COL1A1, LAMA3, PCAT19, GGT5 and SRPX2;

[0036] The calculation formula for the relative coverage of the TSS of each target gene is as follows:

[0037]

[0038] Formula I;

[0039] In the formula, i represents the position of the TSS site of the target gene on the reference genome.

[0040] represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 2, where TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp downstream of the TSS site of the target gene to 599 bp downstream of the TSS site;

[0041] S2, data input module: used to input the sample data into the multiple myeloma minimal residual disease monitoring model;

[0042] The multiple myeloma minimal residual disease monitoring model is prepared according to the following steps: using the TSS relative coverage data of the 50 target genes and the corresponding minimal residual disease conditions of the multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects as training samples to train the multiple myeloma minimal residual disease monitoring model;

[0043] The minimal residual disease status is multiple myeloma minimal residual disease negative or multiple myeloma minimal residual disease positive;

[0044] S3. Result output module: used to output the probability value of whether the multiple myeloma patient to be tested is minimal residual disease positive or minimal residual disease negative through the multiple myeloma minimal residual disease monitoring model; and then derive the minimal residual disease status of the multiple myeloma patient to be tested based on the probability value.

[0045] The computer-readable storage medium mentioned above refers to a carrier for storing data, which may be a tape, disk, floppy disk, CD, magneto-optical disk, ROM, PROM, VCD, DVD, hard disk, flash memory, U disk, CF card, SD card, MMC card, SM card, memory stick (MemoryStick) or xD card, etc.

[0046] In a fifth aspect, the present invention provides a method for monitoring or assisting in monitoring minimal residual disease in a multiple myeloma patient, comprising the following steps:

[0047] S1. Obtain TSS relative coverage data of 50 target genes in patients with multiple myeloma to be tested;

[0048] The 50 target genes are ADGRL2, CNN3, COL11A1, CD244, SDC1, TMEM37, MAP2, CNTN4, ENTPD3, FAM3D, FBXO40, TMEM108, WWTR1, LINC01322, NAALADL2, KCNMB2-AS1, LINC01206, GMNC, SHROOM3, SOWAHB, LRAT, ATP10B, SAMD3, MOXD 1. ANLN, FOXP2, PTPRZ1, TUSC3, LINC01605, KCNQ3, ANO3, MUC15, BARX2, SCNN1A, PTHLH, LRIG3-DT, LINC00 402, RHOV, DLL4, ITPKA, FBN1-DT, ONECUT1, C2CD4A, ISLR2, CHRNB4, COL1A1, LAMA3, PCAT19, GGT5 and SRPX2;

[0049] The calculation formula for the relative coverage of the TSS of each target gene is as follows:

[0050]

[0051] Formula I;

[0052] In the formula, i represents the position of the TSS site of the target gene on the reference genome.

[0053] represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 2, where TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp downstream of the TSS site of the target gene to 599 bp downstream of the TSS site;

[0054] S2. inputting the sample data into a multiple myeloma minimal residual disease monitoring model;

[0055] The multiple myeloma minimal residual disease monitoring model is prepared according to the following steps: using the TSS relative coverage data of the 50 target genes and the corresponding minimal residual disease conditions of the multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects as training samples to train the multiple myeloma minimal residual disease monitoring model;

[0056] The minimal residual disease status is multiple myeloma minimal residual disease negative or multiple myeloma minimal residual disease positive;

[0057] S3. Outputting whether the multiple myeloma patient to be tested is minimal residual disease positive or minimal residual disease negative through the multiple myeloma minimal residual disease monitoring model; and then deriving the minimal residual disease status of the multiple myeloma patient to be tested based on the probability value.

[0058] In a fifth aspect, the present invention provides a method for constructing a multiple myeloma minimal residual disease monitoring model, the method comprising the following steps: using the TSS relative coverage data of the 50 target genes of known multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects and the corresponding minimal residual disease conditions as training samples to train the multiple myeloma minimal residual disease monitoring model.

[0059] The TSS relative coverage data of the 50 target genes of the multiple myeloma patient to be tested is derived from the cfDNA in the test sample of the multiple myeloma patient to be tested;

[0060] The circulating free DNA (cfDNA) in the present invention is the most commonly used liquid biopsy analyte. cfDNA exists in circulating blood and is a DNA fragment derived from cell apoptosis and necrosis and free from cells. Among them, there is DNA from normal cells and DNA from tumor cells (circuLating tumor DNA, ctDNA).

[0061] The sample to be tested can be selected from one or more of plasma, serum, whole blood, separated blood cells, biopsy tissue, tissue sections, paraffin-embedded tissue, body fluids, and cells.

[0062] Furthermore, the sample to be tested is selected from one or more of plasma, serum, whole blood, and separated blood cells.

[0063] Furthermore, the sample to be tested may be plasma.

[0064] The present invention can be used for the detection of liquid samples such as blood, and has the following advantages: first, sampling is simple, which can significantly reduce costs and reduce trauma and risks to patients; second, compared with biopsy examinations that can only examine specific target tissues each time, liquid biopsy can more easily achieve synchronous coverage of tumors; third, liquid biopsy is easy to operate and has a fast detection speed, and samples can be repeatedly obtained for high-frequency dynamic monitoring; finally, malignant tumors are a heterogeneous disease, and liquid biopsy can reflect the entire genomic picture of the tumor, while reducing the deviation caused by tumor heterogeneity to diagnosis, it can also timely reflect the dynamic changes in tumor development.

[0065] The present invention constructs a minimal residual disease monitoring model for multiple myeloma, which can be used to detect the MRD negative or positive status of multiple myeloma patients, with a sensitivity of about 90% and a specificity of about 95%. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 Flowchart for monitoring MRD status for the Minimal Residual Disease Monitoring Model in Multiple Myeloma.

[0067] Figure 2 Flowchart for the construction and testing of a minimal residual disease monitoring model for multiple myeloma.

[0068] Figure 3 is the ROC curve of the model in the training set.

[0069] Figure 4 is the ROC curve of the model in the validation set. DETAILED DESCRIPTION

[0070] Unless otherwise specified, the experimental methods used in the following examples are conventional methods.

[0071] Unless otherwise specified, the materials and reagents used in the following examples can be obtained from commercial sources.

[0072] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the present invention is further explained in conjunction with specific embodiments below, but the following embodiments are only preferred embodiments of the present invention, not all. Based on the embodiments in the implementation mode, other embodiments obtained by those skilled in the art without creative work are all within the scope of protection of the present invention.

[0073] The extraction of cfDNA from the peripheral blood plasma samples of the subjects and the construction of the cfDNA library in the following examples are as follows:

[0074] 10 ml of peripheral blood from the subject was collected using a blood collection tube with cfDNA stabilizer, and plasma was separated as follows: the blood collection tube was centrifuged using a floor centrifuge, the centrifugal temperature was 4 ° C, the centrifugal speed was 2000 g / min, and the centrifugation was performed for 10 minutes. After obtaining the plasma supernatant, it was transferred to a 2 ml EP tube with a pipette, and a second high-speed centrifugation was performed, the centrifugal temperature was 4 ° C, the centrifugal speed was 16000 g / min, and the centrifugation was performed for 10 minutes. Finally, the plasma was separated into a 1.5 ml EP tube. After the plasma was separated from the peripheral blood sample, MagPure Circulating DNA TL Kit (manufacturer: Guangzhou Meiji Biotechnology Co., Ltd., catalog number: IVD5432) was used to extract cfDNA from fresh or frozen plasma samples of the subject. Please refer to the requirements of the instructions for the operation process. 1000 μL of plasma sample should be taken each time for cfDNA extraction. cfDNA sample requirements: cfDNA concentration is not less than 0.1 ng / μL. Qualified cfDNA is used for subsequent cfDNA library construction.

[0075] The kit used for cfDNA library construction was Hieff DNA Library Prep Kit (Cat. No. 13577) was used to obtain the cfDNA library. The library concentration was qualified if it was ≥ 25 ng / μL.

[0076] Example 1. Obtaining the relative coverage of TSS of target parameters for monitoring model of minimal residual disease in multiple myeloma 1. Design and synthesis of target genes and probes for monitoring minimal residual disease in multiple myeloma

[0077] Through the multiple myeloma cell processing module, the chromatin open interval capture module, the target screening module, the target confirmation and verification module, the multiple myeloma cell-specific peaks screened and the specific peaks in the control cells were analyzed for differential peaks, and the multiple myeloma cell-specific chromatin open intervals were screened. Finally, 50 genes for monitoring micro-residual foci of multiple myeloma were obtained as target genes, as shown in Table 1.

[0078] Table 1 shows the start and end positions of 50 target genes in the human reference genome hg38 version

[0079]

[0080]

[0081] 400 target sites shown in Table 2 were designed for the above 50 target genes. Eight RNA probes were designed and synthesized for each target gene. One RNA probe covered one target site, and a total of 400 RNA probes were able to cover and capture 50 target genes. The 400 RNA probes were single-stranded RNAs encoded by the reverse complementary sequences of the 400 target sites shown in Table 2.

[0082] Table 2 shows the start and end positions of 400 targets for 50 target genes in the human reference genome hg38 version

[0083]

[0084]

[0085]

[0086]

[0087]

[0088] In the above table, from left to right, multiple groups of data are composed of chromosomes, the starting position of the target region, and the ending position of the target region, indicating the chromosome where a target point is located and its specific position.

[0089] 2. Probe Capture

[0090] The probe capture targets the cfDNA library of the subject, and the cfDNA library of the subject to be tested is obtained according to the above method. The hybrid capture reagents kit (Cat. No.: 1005101) was used to capture the cfDNA library and obtain the probe capture library.

[0091] Step 1: Library concentration and hybridization reaction preparation

[0092] 1. Add 5 μL cfDNA library (usually 5-12 libraries mixed) and 2 μL NanoBlockers(for ), each hybridization reaction mixed 10 individual libraries (barcode is used to distinguish different samples), each library sampling amount is in the range of 300-500ng, the sampling amount of libraries with the same sequencing data amount is consistent, and the sampling amount of libraries with different sequencing data amounts is sampled according to the sequencing quantity ratio, the library is mixed into a 1.5mL centrifuge tube, and placed in a vacuum concentrator (heated at 60°C) until all the liquid evaporates and is completely dried, and the centrifuge tube is sealed for standby use to obtain a concentrated library to be hybridized.

[0093] 2. Prepare hybridization reaction body (WT-safe-P) in a 200 μL PCR tube, as shown in Table 3. Add 17 μL of the hybridization reaction system shown in Table 3 to the concentrated library to be hybridized, vortex to mix thoroughly, carefully transfer 17 μL of supernatant to a new PCR tube well with a pipette, and then add 20 μL of mineral oil to the PCR tube to prevent liquid evaporation during hybridization incubation to obtain the hybridization reaction system of the library to be hybridized.

[0094] Table 3 shows the hybridization reaction system (WT-safe-P)

[0095]

[0096]

[0097] In the hybridization reaction system, the above 400 RNA probes are mixed in equal molar ratios.

[0098] Step 2: Hybridization overnight

[0099] 1. Place the hybridization reaction system of the library to be hybridized obtained in step (i) in a PCR instrument, start the hybridization reaction program (Nad_H1), set the temperature of the hot cover to 75°C, and control the hybridization incubation at 65°C within 12-24 hours, as shown in Table 4, to obtain the hybridization reaction product.

[0100] Table 4 is the hybridization reaction program (Nad_H1)

[0101]

[0102] Step (III): Magnetic bead capture and elution

[0103] 1. Preheat wash buffer 2 at 65 degrees for 1 hour;

[0104] 2.50 μL streptavidin magnetic beads (trade name: Streptavidin Beads, manufacturer: Nanoda (Nanjing) Biotechnology Co., Ltd., catalog number: 1005101) to a new 1.5 ml EP tube, place on the magnetic stand for 3 minutes, and remove the supernatant;

[0105] 3. Add 200 μL binding buffer, pipette up and down 10 times, place on the magnetic stand for 3 minutes, and remove the supernatant;

[0106] 4. Repeat step 3 on both sides, wash 3 times in total, add 20 μL binding buffer, pipette up and down 10 times, transfer to a 0.2 ml PCR tube, and place in a PCR instrument at 65 degrees for 5 minutes;

[0107] 5. Add the captured magnetic beads incubated at 65 degrees to the hybridization reaction product obtained in step (ii) above, incubate at 65 degrees for 45 minutes, pipette up and down 10 times every 15 minutes; place on the magnetic stand for 3 minutes, and remove the supernatant;

[0108] 6. Add 500 μL wash buffer 2, pipette up and down 10 times, incubate at 65 degrees for 10 minutes, place on the magnetic stand for 3 minutes, and remove the supernatant;

[0109] 7. Repeat step 9 twice, add 23 μL of water to resuspend the magnetic beads to obtain the magnetic bead capture product.

[0110] Step 4: PCR amplification after capture

[0111] 1. Prepare the PCR reaction system (KPMix) in an ice box or on ice. The preparation method is shown in Table 5. Add 50 μL of the PCR reaction system to each magnetic bead capture product, vortex to mix, resuspend the magnetic beads, and centrifuge instantly to obtain the PCR reaction system of the magnetic bead capture product.

[0112] 2. Transfer the PCR reaction system of the magnetic bead-captured product to a new PCR tube and place it in the PCR instrument.

[0113] 3. Start the PCR reaction program (PCP) and set the heated cover to 105°C, see Table 6.

[0114] Table 5 is the PCR reaction system (KPMix)

[0115] Component name Component volume (μL) KAPA HiFi HotStart ReadyMix 25 Primer Mix (NBI-PF / NBI-PR) 5 Nuclease Free Water 20 total 50

[0116] The Primer Mix contains two oligonucleotides complementary to the universal sequence of the probe-enriched library DNA adapter. The specific sequences are as follows:

[0117] The nucleotide sequence of NBI_PF is: AATGATACGGCGACCACCGAGATCTACAC (SEQ ID NO: 1);

[0118] The nucleotide sequence of NBI_PR is: CAAGCAGAAGACGGCATACGAGAT (SEQ ID NO: 2).

[0119] Table 6 shows the post-capture PCR reaction program

[0120]

[0121] Note: The selection of amplification cycle number is based on the coverage of the panel probes and the total amount of library input in the hybridization reaction system. For NAD Probes Pool Name: cfDNA-safe, if the total amount of library input is ≤1μg, the amplification cycle number should be set to 16, otherwise it should be set to 14.

[0122] Step 5: Purification of amplified products

[0123] 1. After the PCR reaction program in step (iv) above is completed, the PCR tube is centrifuged instantly, and the capture library with the lowest total amount of hybridization reaction library input is selected. 1 μL of PCR product is used for concentration determination using the Equalbit 1x dsDNA HS Assay Kit. If the initial DNA concentration of the capture library is ≥5 ng / μL, it is qualified and the following steps can be continued.

[0124] 2. Add 60 μL DNA purification magnetic beads (trade name: VAHTS DNA CleanBeads, manufacturer: Nanjing Novogene Biotech Co., Ltd., item number: N411-03) to each 50 μL PCR product, vortex to mix, centrifuge instantly, incubate at room temperature for 5 minutes, and place the PCR tube on the magnetic stand for 1-2 minutes. After the magnetic beads are completely attached to the wall, discard the supernatant with a pipette.

[0125] 3. Add 200 μL 80% ethanol (prepared on the same day) to rinse the DNA purification magnetic beads. Do not blow on the magnetic beads. Centrifuge briefly to absorb the residual ethanol. Repeat this step once.

[0126] 4. Use a pipette to absorb the remaining ethanol again and place it on the magnetic stand at room temperature for 5-10 minutes until it is dry (the magnetic beads are not reflective, not cracked or slightly cracked).

[0127] 5. Add 20 μL Nuclease-Free Water to the dried DNA purification magnetic beads for elution, vortex to mix thoroughly, place at room temperature for 2 minutes, centrifuge briefly and place on a magnetic stand, let stand for 2 minutes until the magnetic beads are completely attached to the wall, then use a pipette to carefully transfer 19 μL of supernatant to the 0.2 mL single tube prepared in advance to obtain the capture library.

[0128] 6. The concentration of the captured library was determined using the Qubit 4.0 fluorescence quantifier and the Equalbit 1x dsDNA HS Assay Kit reagent. The quality inspection standard for the concentration of the captured library was: library concentration ≥ 5 ng / μL; DNA library volume ≥ 20 μL.

[0129] 3. NGS Next Generation Sequencing

[0130] The captured library of the subject obtained above was subjected to NGS second-generation sequencing to obtain sequencing data, which was processed as follows:

[0131] 1. Sequencing data preprocessing and calculation

[0132] 1) Data filtering of sequencing data:

[0133] The original sequencing sequence obtained by sequencing contains low-quality reads with adapters. In order to ensure the quality of information analysis, it is necessary to finely filter the raw reads to obtain clean reads, and subsequent analysis is based on clean reads. The steps of data filtering are as follows:

[0134] (1) Remove the sequence with adapter in the read;

[0135] (2) If the number of unknown bases (N) in a read exceeds 5, the entire read is filtered out;

[0136] (3) Trim the bases with 3'-end quality values ​​lower than 20 in the read until the quality value is greater than 20;

[0137] (4) truncate the first 3 bases at the 5' end of the read;

[0138] (5) Remove reads with sequence length less than 35 bp after the above steps;

[0139] (6) If read1 and read2 undergo the above steps and only read1 (or read2) remains and cannot form a pair of reads, the read is discarded.

[0140] The sequencing adapter sequence information of DNA-Seq is as follows (the underlined part represents the 8-bit index sequence):

[0141] P5 Adapter:5'–AATGATACGGCGACCACCGAGATCTACAC NNNNNNNN ACACTCTTTCCCTACACGACCGTCTCTCCGATCT-3' (SEQ ID NO: 3),

[0142] P7 Adapter:

[0143] 5'-GATCGGAAGAGCACACGTCTGAACTCCAGTCAC NNNNNNNN ATCTCGTATGCCGTCTTCTGCTTG-3' (SEQ ID NO: 4).

[0144] 2) Data quality control of sequencing data:

[0145] Base quality value and error rate are important indicators for measuring sequencing quality. The higher the quality value, the lower the probability of base errors. The sequencing error rate of each base can be calculated by the formula Qphred = -10log 10 (e) is converted, where e is the base sequencing error rate, and the Qphred value is calculated through a probability model during the Base Calling process, which can accurately predict the error rate of base discrimination.

[0146] The base error rate is affected by multiple factors including the sequencer itself, sequencing reagents, and samples. For the Illumina high-throughput sequencing platform, the sequencing quality value distribution has two characteristics:

[0147] (1) The sequencing quality value will decrease as the sequencing progresses. This is because factors such as incomplete cleavage of fluorescent markers during the sequencing process will cause the fluorescent signal to attenuate, thereby increasing the error rate and decreasing the quality value.

[0148] (2) The first few bases of each read will also have a higher sequencing error rate and a relatively low quality value. This is because at the beginning of the sequencing-by-synthesis process, the fluorescent photosensitive element of the sequencer focuses slowly and the quality of the acquired fluorescent image is low, resulting in a higher base recognition error rate and a lower quality value.

[0149] The data GC content distribution check is used to detect the presence of AT and GC separation, which may be caused by sequencing or library construction and will affect subsequent quantitative analysis. In theory, the content of G and C bases and A and T bases in a common library should be equal in each sequencing cycle, and the entire sequencing process should be stable and unchanged, showing a horizontal line. In reality, due to reasons such as random primer amplification deviation and sequencing technology, there are often large fluctuations in the first 6-7 bases and the last few bases of each read obtained by sequencing. This fluctuation is normal.

[0150] 2. Sequencing data comparison and analysis

[0151] Cleandata was aligned to the reference genome GRCh38 through BWA to obtain the initial alignment results in SAM format. Samtools was used to rearrange the alignment results, create indexes, remove duplicates, low coverage regions and genomic blacklist regions, and obtain the final alignment result file in BAM format. Finally, the chromosomes were divided into adjacent, non-intersecting window areas, and the number and proportion of long and short cfDNA in each window area were counted, and the statistical numbers were corrected and processed for GC content.

[0152] Finally, the sequencing data were compared and analyzed to obtain the BAM file of the probe capture sequencing data (containing the sequencing data of 50 target genes).

[0153] 4. Obtaining the relative coverage of TSS of target genes

[0154] Based on the human TSS (transcription start site, counted as +1bp) position information provided by the human reference genome hg38 version, the TSS sites in 50 target genes were defined. The TSS site is also the central position of the target site (eight probes were designed for each target gene to cover the target site).

[0155] Taking the TSS site of each target gene as the center, the Samtools tool was used to calculate the sequencing depth of the TSS region of each target gene spanning 100 base pairs upstream and downstream of the TSS according to the probe capture sequencing data BAM file obtained in the above three steps; at the same time, the Samtools tool was also used to calculate the sequencing depth in the upstream (-600bp to -400bp) and downstream (400bp-600bp) regions of the TSS site, which was defined as the background value of the target point.

[0156] The TSS relative coverage of each target gene was calculated according to the following formula I:

[0157]

[0158] In the formula, i represents the position of the TSS site of the target gene on the reference genome.

[0159] represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site. represents the sum of the sequencing depths of 200 base pairs of TSS region 2, wherein TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site, It represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp base pairs downstream of the TSS site of the target gene to 599 bp base pairs downstream of the TSS site.

[0160] Example 2: Construction of a monitoring model for minimal residual disease in multiple myeloma

[0161] Figure 1 Flowchart for monitoring MRD status for the Minimal Residual Disease Monitoring Model in Multiple Myeloma. Figure 2 Flowchart for the construction and testing of a minimal residual disease monitoring model for multiple myeloma.

[0162] 1. Inclusion Criteria

[0163] A total of 100 multiple myeloma MRD-positive samples confirmed by clinical pathology and 100 multiple myeloma MRD-negative samples by clinical pathology were enrolled, and the sample type was peripheral blood.

[0164] The method of clinical pathological detection of MRD is to recruit paired bone marrow blood from patients and determine whether there are micro-residual lesions by bone marrow biopsy immunohistochemistry. The basis of the determination method: In 2016, the International Myeloma Working Group (IMWG) defined the presence of MRD in myeloma as one tumor cell among at least 10-5 normal cells in the bone marrow (BM) (the minimum sensitivity threshold is 10 -5 ), whose detection is considered an important endpoint.

[0165] Inclusion criteria for the study subjects: 1. Chinese patients aged 18-80 years old, regardless of gender; 2. Multiple myeloma patients who have entered remission after treatment; 3. No previous combination of other malignant tumors.

[0166] The samples were divided into a training set (70% of the original data set) and a validation set (30% of the original data set).

[0167] 2. Construction of a monitoring model for minimal residual disease in multiple myeloma

[0168] 1. Obtaining TSS relative coverage data of target genes

[0169] 100 MRD-positive samples and 100 MRD-negative samples of multiple myeloma were tested as samples for the following tests:

[0170] cfDNA was extracted from 200 peripheral blood plasma samples and a library was constructed according to the above method to obtain a cfDNA library; probe capture was then performed according to the method in step 2 of Example 1 to obtain a captured library, and sequencing was then performed according to the method in step 3 of Example 1 to obtain sequencing data, and TSS relative coverage was calculated according to the method in step 4 of Example 1 to obtain TSS relative coverage of 50 target genes for each sample to be tested.

[0171] 2. Construction of a monitoring model for minimal residual disease in multiple myeloma

[0172] The TSS relative coverage of the 50 target genes and the corresponding MRD results (MRD positive or MRD negative) of each sample to be tested were converted into a data matrix, and a total of 200 data matrices of samples were obtained (for example, Table 7), of which the data matrices of 70 MRD-positive samples and 70 MRD-negative samples were used as model training data sets.

[0173] Table 7 shows the data matrix of 6 samples in the training data set of the example model

[0174]

[0175]

[0176] Note: T1-T50 represent the TSS of 50 target genes, MRD represents the result of clinical pathological detection of MRD, 1 represents positive, 0 represents negative, S1-S6 represent 6 samples.

[0177] The model training data set containing the TSS relative coverage of 50 target genes and MRD results of 140 samples was used as the input data for modeling. The R package 'caret' was used to execute the machine learning algorithm that combined the random forest algorithm and the gradient boosting machine algorithm to establish a multiple myeloma minimal residual disease monitoring model; the model training was iteratively optimized using the three-repeated ten-fold cross-validation method. The optimal parameters of the model are as follows: the number of trees is 1500, the maximum depth of each tree is 15, the maximum number of features considered for each split is auto, the minimum number of samples that a node can continue to split is 5, and the minimum number of samples required for a leaf node is 3.

[0178] According to the probability value of each sample in the training set output by the multiple myeloma minimal residual disease monitoring model belonging to the MRD positive and negative categories, different thresholds are used to convert the probability into the predicted category of MRD positive or negative, and the ROC curve is drawn ( Figure 3 ), calculated the AUC of the model in the training set = 0.975, calculated the Precision and Recall at each threshold, drew the Precision-Recall curve, and determined that the reasonable threshold of the model was 0.53. If the model prediction probability value of the sample was greater than 0.53, the sample was judged to be MRD positive. If the model prediction probability value of the sample was less than or equal to 0.53, the sample was judged to be MRD negative. The specificity of the model in the training set was 94.5%, and the sensitivity was 87.2%.

[0179] 3. Validation set ROC curve and AUC value evaluate the overall performance of the model

[0180] The data matrix of the remaining 30 MRD-positive and 30 MRD-negative samples in the data matrix of the 200 samples obtained in 2 above was used as the validation set. The validation set data was input into the above-mentioned multiple myeloma minimal residual disease monitoring model to obtain the model prediction probability value and prediction category. Then, according to the clinical pathological MRD results of the validation set samples, the ROC curve was drawn ( Figure 4 ), the AUC of the calculated model in the validation set is 0.965.

[0181] The confusion matrix can be used to more clearly evaluate the accuracy, specificity and sensitivity of the model (Table 8). The confusion matrix divides the true labels and model prediction results into four categories: true positive (TP), false positive (FP), true negative (TN) and false negative (FN).

[0182] Table 8 shows the confusion matrix of the model prediction results in the validation set

[0183]

[0184] According to the results of the confusion matrix, the accuracy, specificity and sensitivity are calculated to analyze the performance of the model. The formula for calculating accuracy and specificity is as follows:

[0185] Accuracy = (TP+TN) / (TP+TN+FP+FN)

[0186] Specificity = TN / (TN+FP)

[0187] Sensitivity = TP / (TP+FN)

[0188] Ultimately, the accuracy of the multiple myeloma minimal residual disease monitoring model in the validation set was 90%, the specificity was 93.3%, and the sensitivity was 86.6%.

[0189] Example 3: Multiple Myeloma Minimal Residual Disease Monitoring Model Detects Real Clinical Samples

[0190] Another 50 Chinese patients aged 18-80 years old, regardless of gender, with multiple myeloma who entered remission after treatment and had no history of other malignant tumors were independently enrolled, of which 31 were MRD-negative by clinical pathology and 19 were MRD-positive by clinical pathology. These 50 samples were used as independent, real clinical samples to evaluate the real clinical performance and generalization performance of the model.

[0191] cfDNA was extracted from 50 samples to be tested (patients' peripheral blood plasma) and a library was constructed according to the above method to obtain a cfDNA library; probe capture was then performed according to the method in step 2 of Example 1 to obtain a captured library, and sequencing was then performed according to the method in step 3 of Example 1 to obtain sequencing data, and TSS relative coverage was calculated according to the method in step 4 of Example 1 to obtain TSS relative coverage data of 50 target genes of each sample to be tested.

[0192] The TSS relative coverage of the 50 target genes of each sample to be tested and the corresponding MRD results (MRD positive or MRD negative) are converted into a data matrix, and a data matrix of 50 samples is obtained in total. The data matrix of 50 samples is input into the multiple myeloma minimal residual disease monitoring model prepared in Example 2 to obtain the model prediction probability value and prediction category.

[0193] If the model prediction probability value of the sample is greater than 0.53, the sample is judged to be MRD-positive, and if the model prediction probability value of the sample is less than or equal to 0.53, the sample is judged to be MRD-negative.

[0194] The results of the model prediction are: 19 true positives were detected by the model (TP=19), 29 true negative samples were correctly detected (TN=29), and 2 clinically determined multiple myeloma MRD negative samples were incorrectly determined as positive by the model (FP=2). No positive samples were incorrectly classified as negative (FN=0). Therefore, the model of the method of the present invention has a prediction accuracy of 96%, a specificity of 93.5%, and a sensitivity of 100% in these 50 samples as independent, real clinical samples.

[0195] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the following steps: S1. Receive data: Receive sample data, wherein the sample data is TSS relative coverage data of 50 target genes of a multiple myeloma patient to be tested; The 50 target genes are ADGRL2, CNN3, COL11A1, CD244, SDC1, TMEM37, MAP2, CNTN4, ENTPD3, FAM3D, FBXO40, TMEM108, WWTR1, LINC01322, NAALADL2, KCNMB2-AS1, LINC01206, GMNC, SHROOM3, SOWAHB, LRAT, ATP10B, SAMD3, MOXD 1. ANLN, FOXP2, PTPRZ1, TUSC3, LINC01605, KCNQ3, ANO3, MUC15, BARX2, SCNN1A, PTHLH, LRIG3-DT, LINC00 402, RHOV, DLL4, ITPKA, FBN1-DT, ONECUT1, C2CD4A, ISLR2, CHRNB4, COL1A1, LAMA3, PCAT19, GGT5 and SRPX2; The calculation formula for the TSS relative coverage of each of the 50 target genes is as follows: Formula I; Among them, the formula Represents the location of the TSS site of the target gene on the reference genome, represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 2, where TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp downstream of the TSS site of the target gene to 599 bp downstream of the TSS site; S2. Input data: input the sample data into the multiple myeloma minimal residual disease monitoring model; The multiple myeloma minimal residual disease monitoring model is constructed according to the following steps: using the TSS relative coverage data of the 50 target genes of the multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects and the corresponding minimal residual disease conditions as training samples to train the multiple myeloma minimal residual disease monitoring model; The minimal residual disease status is multiple myeloma minimal residual disease negative or multiple myeloma minimal residual disease positive; S3. Output result: output the probability value of whether the multiple myeloma patient to be tested is minimal residual disease positive or minimal residual disease negative through the multiple myeloma minimal residual disease monitoring model; and then derive the minimal residual disease status of the multiple myeloma patient to be tested based on the probability value.

2. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps in the computer device of claim 1 are implemented.

3. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program causes a computer to execute the steps in the computer device according to claim 1.

4. A device for monitoring minimal residual disease in multiple myeloma, comprising: S1, data receiving module: used to receive sample data, wherein the sample data is TSS relative coverage data of 50 target genes of multiple myeloma patients to be tested; The 50 target genes are ADGRL2, CNN3, COL11A1, CD244, SDC1, TMEM37, MAP2, CNTN4, ENTPD3, FAM3D, FBXO40, TMEM108, WWTR1, LINC01322, NAALADL2, KCNMB2-AS1, LINC01206, GMNC, SHROOM3, SOWAHB, LRAT, ATP10B, SAMD3, MOXD 1. ANLN, FOXP2, PTPRZ1, TUSC3, LINC01605, KCNQ3, ANO3, MUC15, BARX2, SCNN1A, PTHLH, LRIG3-DT, LINC00 402, RHOV, DLL4, ITPKA, FBN1-DT, ONECUT1, C2CD4A, ISLR2, CHRNB4, COL1A1, LAMA3, PCAT19, GGT5 and SRPX2; The calculation formula for the relative coverage of the TSS of each target gene is as follows: Formula I; Among them, the formula Represents the location of the TSS site of the target gene on the reference genome, represents the sum of the sequencing depths of 200 base pairs of TSS region 1, where TSS region 1 is a region consisting of 200 base pairs from 100 base pairs upstream of the TSS site of the target gene to 99 base pairs downstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 2, where TSS region 2 is a region consisting of 200 base pairs from 600 bp upstream of the TSS site of the target gene to 401 bp upstream of the TSS site; represents the sum of the sequencing depths of 200 base pairs of TSS region 3, where TSS region 3 is a region consisting of 200 base pairs from 400 bp downstream of the TSS site of the target gene to 599 bp downstream of the TSS site; S2, data input module: used to input the sample data into the multiple myeloma minimal residual disease monitoring model; The multiple myeloma minimal residual disease monitoring model is prepared according to the following steps: using the TSS relative coverage data of the 50 target genes and the corresponding minimal residual disease conditions of the multiple myeloma minimal residual disease-negative and multiple myeloma minimal residual disease-positive subjects as training samples to train the multiple myeloma minimal residual disease monitoring model; The minimal residual disease status is multiple myeloma minimal residual disease negative or multiple myeloma minimal residual disease positive; S3. Result output module: used to output the probability value of whether the multiple myeloma patient to be tested is minimal residual disease positive or minimal residual disease negative through the multiple myeloma minimal residual disease monitoring model; and then derive the minimal residual disease status of the multiple myeloma patient to be tested based on the probability value.

Images