A data processing method for determining a DNA nucleic acid organism by a biological nanopore

Abstract

The application discloses a data processing method for determining DNA nucleic acid organisms by using a biological nanopore, which comprises a series of key steps such as data correction, noise filtering, event screening, level searching and sequence alignment from the collection of original current data to the generation of final DNA base sequence and the determination of organisms, and constructs a complete data processing flow, realizes comprehensive processing and analysis of the biological nanopore current signal, and can realize real-time acquisition of DNA sequence information in a way of collecting and analyzing simultaneously, thereby greatly improving the detection efficiency. The method does not need to wait until all data collection is completed before analysis, thereby saving a large amount of time, and the biological nanopore is used as a detection means, the sample preparation requirement is low, and complex sample processing is not needed, and the data analysis flow is clear and understandable.

Description

Technical Field

[0001] This invention belongs to the field of bioinformatics nanopore sequencing technology, specifically relating to a data processing method for identifying DNA and nucleic acid organisms using biological nanopores. Background Technology

[0002] Bio-nanopores are self-assembling protein pores. When these pores are embedded in phospholipid bilayers or polymer membranes, current can be detected by adding an ionization buffer and applying a suitable voltage. When a analyte is added to the system, a blocking current can be detected as the analyte passes through. Theoretically, the blocking current data represents the current signal generated by the detected analyte. However, other abnormal events caused by environmental or experimental factors cannot be ruled out, although such occurrences are generally rare.

[0003] The principle behind this experiment and detection is well-known in the industry, but only Oxford Nanopore in the UK has mastered and commercially applied this technology. In China, there is still significant room for exploration in experimental techniques, application development, and data processing. The maturation of these techniques will advance the development of gene sequencing applications both domestically and internationally.

[0004] Bio-nanopore technology enables rapid and accurate DNA sequencing, accelerating the DNA sequencing process in genomics research and helping scientists better understand genome structure, function, and variation. In medical diagnostics, this method can be used to detect gene mutations, diagnose genetic diseases, and personalize treatments. Bio-nanopore technology can help scientists better understand the interaction between drugs and target DNA, thus accelerating the drug development process. Furthermore, this technology can be used to detect drug purity and quality and monitor DNA impurities in the biopharmaceutical process. Bio-nanopore technology can be used to detect microbial DNA in the environment, helping to monitor environmental pollution, predict disease outbreaks, and assess the health of ecosystems. In food safety, this technology can be used to detect pathogenic microorganisms or genetically modified ingredients in food, ensuring food quality and safety. As bio-nanopore technology advances, the cost of personal genome sequencing is gradually decreasing, which will provide more possibilities for personalized health management. Individuals can regularly sequence their own genome to understand their genetic characteristics, disease risks, and drug responses, thereby taking appropriate preventative and treatment measures to improve health and quality of life. Bio-nanopore technology can be used in forensic medicine to help detect and identify DNA evidence left at crime scenes, thus helping to solve criminal cases and ensure judicial fairness. Compared to traditional DNA analysis techniques, bio-nanopore technology offers higher resolution and sensitivity, providing more reliable evidence. With the development of application technologies, the use of bio-nanopores for the detection of DNA and nucleic acids in organisms is becoming increasingly widespread. However, the algorithms and implementation schemes for data processing accompanying this technology remain unclear, and data processing methods urgently need to be developed to accelerate the advancement of bio-nanopore-based biological detection technologies. Summary of the Invention

[0005] To address the shortcomings and problems of existing DNA and nucleic acid organism detection methods, this invention provides a data processing method for determining DNA and nucleic acid organisms using biological nanopores. This method aims to improve sequencing accuracy and reliability, simplify the data analysis process, and accelerate the optimization and promotion of nanopore sequencing technology.

[0006] The solution adopted by this invention to solve its technical problem is: a data processing method for identifying DNA and nucleic acid organisms using biological nanopores, comprising the following steps:

[0007] Step 1: Based on the characteristics of bio-nanopores, the DNA sequence to be detected is divided into combinations of several bases, and then the DNA sequence to be detected is detected by bio-nanopores; the raw current change data of the DNA sequence to be detected is obtained.

[0008] Step two: Correct invalid values ​​in the original current change data using a correction algorithm;

[0009] Step 3: Perform high-frequency noise filtering on the corrected current change data;

[0010] Step 4: Divide the processed current change data into orifice current and blockage current, and filter out events that meet the criteria.

[0011] Step 5: Perform level search and fitting for the events that meet the search criteria; first, identify the level, determine the boundary of the current resistance level generated by different base combinations, and then remove false levels.

[0012] Step 6: Establish a reference level database. Select DNA samples with known sequences as the test set, perform nanopore sequencing and level extraction on them, compare the level data of the test set samples with the reference level database to obtain annotated DNA sequences, compare the annotated DNA sequences with the known actual sequences of the test set samples, calculate sequence similarity, analyze the type and origin of mismatched bases, evaluate the accuracy of the reference database, and optimize and improve the reference database based on the evaluation results; such as removing erroneous data and adding new samples.

[0013] Step 7: Based on the Needleman-Wunsch and Smith-Waterman sequence alignment algorithms, the level data obtained from the sequencing of the DNA to be tested in Step 5 is compared with the level data in the reference level database. The most similar sequence is found, the DNA sequence of the DNA to be tested is inferred, the level of the newly measured DNA sequence is added to the reference level database, and the median of different sequence combinations is recalculated to update and improve the reference level database and improve its accuracy.

[0014] In step one, the DNA sequence to be detected is sampled at a frequency of 500k using a biological nanopore.

[0015] In step two, invalid values ​​are those that exceed a preset percentage of normal values, and the correction algorithm includes one of median filtering and weighted average algorithm.

[0016] In step three, a Bayesian low-pass filter is used to filter high-frequency noise from the corrected original current variation data. The sampling frequency in this project is 500 kHz, the filter uses a 4th-order Bessel filter, and the cutoff frequency is 50 kHz (i.e., Fs / 10).

[0017] In step four, "event" refers to a blocking current event generated when a DNA pore is formed. The screening of events includes the following steps:

[0018] Step 1: First, based on a broad threshold detection, select an event duration Time>1S for extraction, and a current range between (0.1, 0.75) of the detected orifice current.

[0019] Step 2: Initialize the local aperture current value based on the number of detected events (L events) and the detected aperture current (L events initialize the local value).

[0020] Step 3, then based on each local open state (L i ) and each Event(L i We will analyze each event individually and further select suitable events.

[0021] The event is further filtered using the following steps:

[0022] Step 1: Detect the before and after states of each numbered event in the event: event_before and event_after, which represent the open states before and after the event, respectively, with an initial value of NaN.

[0023] Step 2: If event is not the first state, calculate the median value of the open states before event as event_before; if event is not the last state, calculate the median value of the state after event as event_after.

[0024] Step 3: If the current event number is greater than 2, and data exists in the previous two open states, then try to fit the data of the previous state using an exponential decay function and update it to event_before.

[0025] Step 4: If the current event number is less than the open state number and the current state has data, try to fit the data of the next state using the exponential decay function and update it to event_after.

[0026] Step 5, exception handling: If event_before and event_after deviate abnormally from the 3sigma value of the open-state current detected by event, they are also set to NaN.

[0027] The establishment of the reference level database in step six includes the following steps:

[0028] Step 1: Select DNA samples with known sequences for nanopore sequencing to obtain raw current data;

[0029] Step 2: Process and analyze the current data to extract reliable level information;

[0030] Step 3: Repeat steps 1-2 multiple times to accumulate level data and establish a reference level database.

[0031] Step 4: Associate the levels in the reference level database with their corresponding known DNA sequences to form a "level-sequence" mapping relationship.

[0032] The beneficial effects of this invention: The data processing method for identifying DNA and nucleic acid organisms using biological nanopores provided by this invention has the following beneficial effects:

[0033] 1. From the acquisition of raw current data to the generation of the final DNA base sequence, a series of key steps were covered, including data correction, noise filtering, event screening, level lookup, and sequence alignment. A complete data processing workflow was constructed, enabling comprehensive processing and analysis of biological nanopore current signals.

[0034] 2. By employing a simultaneous data acquisition and analysis approach, DNA sequence information can be obtained in real time, significantly improving detection efficiency. Analysis is not required until all data acquisition is complete, saving considerable time.

[0035] 3. Using biological nanopores as a detection method, sample preparation requirements are low, no complicated sample processing is required, and the data analysis process is clear and straightforward.

[0036] 4. By setting appropriate thresholds and using statistical methods to identify valid events, the interference of spurious events was eliminated. Level fitting fully considers frameshift conditions, allowing for jumps and backswings during sequence alignment, thus improving the accuracy of sequence generation.

[0037] 5. By continuously improving the database through level extraction of known sequence samples and constantly supplementing and updating the newly measured levels, a dynamic sequence-level association is formed, providing a reliable reference for sequence inference of unknown DNA samples. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the current change data preprocessing of the present invention.

[0039] Figure 2 This is a schematic diagram illustrating the classification of opening current and blocking current in this invention.

[0040] Figure 3 This is a schematic diagram of the event generation of the present invention.

[0041] Figure 4 This is a schematic diagram of the data window traversal of the present invention.

[0042] Figure 5This is a schematic diagram of the recursive retrieval and search for the optimal t2 split point in this invention.

[0043] Figure 6 This is a schematic diagram showing whether the window of this invention has a transition.

[0044] Figure 7 This is a schematic diagram of the optimal transition position of the present invention.

[0045] Figure 8 This is a schematic diagram of the Level fitting results of this invention.

[0046] Figure 9 This is a schematic diagram of the optimal comparison logic of the present invention.

[0047] Figure 10 This is a schematic diagram of the optimal route of the present invention. Detailed Implementation

[0048] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0049] Example 1

[0050] This embodiment provides a data processing method for identifying DNA and nucleic acid organisms using biological nanopores, characterized by the following steps:

[0051] Step 1: Based on the characteristics of the bio-nanopore, the DNA sequence to be detected is divided into combinations of several bases. For example, the DNA sequence to be detected can be designed as a group of 4 bases, 5 bases, or 6 bases, etc., resulting in sequence combinations of 44, 45, 46, etc. Then, the DNA sequence to be detected is subjected to bio-nanopore detection to obtain the raw current change data of the DNA sequence to be detected. The DNA sequence to be detected is sampled at a frequency of 500k for bio-nanopore detection.

[0052] Step 2: Correct invalid values ​​in the original current change data using a correction algorithm. Invalid values ​​are those that exceed the preset percentage of the normal value. For example, values ​​that exceed the preset 30% of the normal value are invalid. The correction algorithm includes one of median filtering or weighted average algorithm.

[0053] Step 3: Perform high-frequency noise filtering on the corrected current change data, such as... Figure 1 As shown, a Bayesian low-pass filter is used to filter high-frequency noise from the corrected original current variation data. The sampling frequency in this project is 500 kHz, the filter uses a 4th order Bessel filter, and the cutoff frequency is 50 kHz (i.e., Fs / 10).

[0054] Step four, as Figure 2As shown, the processed current change data is divided into open-pore current and blocking current, and the states at this time are called open state and blocked state. Events that meet the criteria are then filtered; an event refers to a blocking current event generated when DNA passes through a pore, such as... Figure 3 The image shows the generation of an event. The filtering of events includes the following steps:

[0055] Step 1: First, based on a broad threshold detection, select an event duration Time>1S for extraction, and a current range between (0.1, 0.75) of the detected orifice current.

[0056] Step 2: Initialize the local aperture current value based on the number of detected events (L events) and the detected aperture current (L events initialize the local value).

[0057] Step 3, then based on each local open state (L i ) and each Event(L i We will analyze each event individually and further select suitable events.

[0058] Further filtering of events includes the following steps:

[0059] Step 1: Detect the before and after states of each numbered event in the event: event_before and event_after, which represent the open states before and after the event, respectively, with an initial value of NaN.

[0060] Step 2: If event is not the first state, calculate the median value of the open states before event as event_before; if event is not the last state, calculate the median value of the state after event as event_after.

[0061] Step 3: If the current event number is greater than 2, and data exists in the previous two open states, then try to fit the data of the previous state using an exponential decay function and update it to event_before.

[0062] Step 4: If the current event number is less than the open state number and the current state has data, try to fit the data of the next state using the exponential decay function and update it to event_after.

[0063] Step 5, exception handling: If event_before and event_after deviate abnormally from the 3sigma value of the open-state current detected by event, they are also set to NaN.

[0064] Step 5: Perform level search and fitting on the events that meet the search criteria. The level search includes the following steps:

[0065] Step 1: Perform a level lookup on valid event data; first, define the data window. Traversal The entire event, such as Figure 4 As shown.

[0066] Step 2: Next, define t1 and t3 as window time positions, and recursively search for the optimal t2 split point, such as... Figure 5 As shown, the evaluation criterion is that the smaller the value, the better; see Formula 3 in the instruction manual for details.

[0067] Step 3: Then, search for t2 again in both windows (t1, t2) and (t2, t3). If the window has no transitions, expand the search range; if the window has multiple transitions, narrow the search range, such as... Figure 6 As shown.

[0068] Step 4: Finally, by repeatedly performing steps 2 and 3, all optimal transition positions are found, and the results are as follows. Figure 7 It can search for all levels in an event. Of course, when evaluating the optimal transition position, the following condition must be met: P(t1,t2,t3) is less than the set threshold. See Formula 3 in the manual for details.

[0069] The detailed methods and formulas involved in level lookup mainly involve two steps: first, identifying the level and determining its boundaries within the time trajectory; and second, removing false levels. The specific steps are as follows:

[0070] Step 1: Determine the level location. First, examine a portion of the current trajectory and divide this portion into two parts. Assuming the sampled current within each level follows a Gaussian distribution, calculate the total probability that the two current segments originate from two different Gaussian distributions. Divide this total probability by the probability of the null hypothesis, i.e., the combination of the two parts originating from a single Gaussian distribution. For ease of calculation, logarithmic probability is used. For a given window of current, the average observed current value is I. mean The width is w, and for this part, the single current I within time T. T At time t, the log probability (density) is:

[0071]

[0072] Step 2, identify the observations between T1 and T2 that belong to I. mean The total log probability of a single level, defined by w, is summed from the formulas, i.e., Formula 1 between T1 and T2, and given by defining const using sigma:

[0073] LogP(I t1，t2 )=(t2-t1)Logo(t2,t1)+const Formula 2

[0074] in:

[0075]

[0076] Step 3: Calculate the total log probability, one of the two parts between T1 and T2, and between T2 and T3. To compare the log probabilities, subtract the total log probability of the null hypothesis from the calculated equation, combine all probabilities to produce a comparison metric, and express it as LogP(T1,T2,T3):

[0077] LogP(t1,t2,t3)=(t2-t1)Logσ(t2,t1)+(t3-t2)Logσ(t3,t2)-(t3-t1)Logσ(t3,t1) Formula 3

[0078] in:

[0079]

[0080] The minimum LogP(t1,t2,t3) generated by T2 represents the position where a level transition may occur between the currently observed t1 and t3. In the level search, starting from a given time window [T1,T3], the search minimizes LogP at T2. If min(LogP) is less than a specified threshold (LogP = -50), a level transition exists at T2. If min(LogP) is higher than the threshold of the original time window, no transition occurs between t1 and t3, and increasing the value of t3 is considered to expand the window to a larger value.

[0081] Step 4: Delete false levels. Not all levels meet the requirements. False levels are generated due to other reasons, such as DNA blockage, environmental disturbances, etc. Delete levels with a duration of less than 500uS, an average value exceeding the threshold range [9-72pA], or an average value error greater than 5pA. Finally, use the logarithm calculated by the formula. In addition, similar levels will be identified in Formula 3 and their levels will be merged.

[0082] Step six: Establish a reference level database. After identifying the event of the DNA to be detected, the event is searched for its level. Levels are generated due to different base combinations. For example, each time a bio-nanopore reads four bases, different combinations of the 256 bases (AAAA, AAAC, AAAG, etc.) produce essentially different levels, mainly reflected in the amplitude and duration of current blocking. Figure 8 As shown, a standard level reference database is constructed. DNA samples with known sequences are selected as the test set, and nanopore sequencing and level extraction are performed on them. The level data of the test set samples are compared with the reference level database to obtain annotated DNA sequences. The annotated DNA sequences are then compared with the known actual sequences of the test set samples to calculate sequence similarity, identify the type and origin of mismatched bases, and evaluate the accuracy of the reference database. Based on the evaluation results, the reference database is optimized and improved, such as removing erroneous data and adding new samples.

[0083] The detailed method for establishing a reference level database includes the following steps:

[0084] Step 1: Select DNA samples with known sequences for nanopore sequencing to obtain raw current data;

[0085] Step 2: Process and analyze the current data to extract reliable level information;

[0086] Step 3: Repeat steps 1-2 multiple times to accumulate level data and establish a reference level database.

[0087] Step 4: Associate the levels in the reference level database with their corresponding known DNA sequences to form a "level-sequence" mapping relationship.

[0088] Step 7: Based on the Needleman-Wunsch and Smith-Waterman sequence alignment algorithms, the level data obtained from the sequencing of the DNA to be tested in Step 5 is compared with the level data in the reference level database. The most similar sequence is found, the DNA sequence of the DNA to be tested is inferred, the level of the newly measured DNA sequence is added to the reference level database, and the median of different sequence combinations is recalculated to update and improve the reference level database and improve its accuracy.

[0089] The alignment algorithm uses Needleman-Wunsch and Smith-Waterman sequence alignment algorithms. The sequence from the database is designated A, and the actual sequence is designated B. Needleman-Wunsch or Smith-Waterman alignment of two base sequences A and B allows for gaps between the two sequences. Due to possible gaps, the optimal alignment between the first nA base of A and the first nb base of B is one of the following:

[0090] The comparison and scoring formula is as follows:

[0091]

[0092] in:

[0093]

[0094] Path optimization:

[0095]

[0096] 1. The first Aseq of A i-1 All Bseq bases and B j The optimal alignment between bases plus the gaps in A;

[0097] 2. All Aseq of A i The first Bseq of bases and B j-1 The optimal alignment between bases plus the gaps in B;

[0098] 3. The first Aseq of sequence A in the two sequences. i-1 The first Bseq of bases and B j-1 The optimal alignment between the bases plus the final matching bases (within a window of 20) (as shown in Equation 4) is implemented as follows: Figure 9 As shown. The longer optimal alignment is recursively calculated based on the shorter optimal sub-alignment, and the entries in the alignment table are filled from the top left to the bottom right corner as follows. Figure 10 As shown.

[0099] The level sequence can sometimes go backwards. If the Needleman-Wunsch scheme is used directly, each entry in the alignment table will depend on the entry on the left due to the forward step and the entry on the right due to the backward step, which in turn will depend on the entry in question.

[0100] To address this issue, each step in the alignment is required to advance sequence A by one step. Therefore, the alignment is the optimal mapping of each level in sequence A to its corresponding level in sequence B, or, if no good match exists, an empty level. Thus, the alignment tracking will proceed through levels skipped in sequence B, rather than skipping them.

[0101] A) The Needleman-Wunsch alignment takes into account the horizontal and vertical steps in the alignment table with a penalty w, corresponding to a mismatched base in one or the other sequence.

[0102] A diagonal step indicates a matching base and usually incurs no penalty unless there is a mismatch.

[0103] B) Our nanopore arrangement forces each step forward along sequence A, but allows backward movement in sequence B. Affine penalties are assigned to jumps and backward movement: for example, backward movement of 3 levels will result in a backward penalty wback plus twice the backward penalty wback (as in Equation 5).

[0104] C) Differences between Needleman-Wunsch alignment and nanopore alignment when there are jumps in sequence B.

[0105] D) Following this process (through a global optimal mechanism for tracing back to the source, as shown in Formula 6), we will perform biological nanopore detection on the DNA sequence of the unknown species, obtain this level, compare it with the database, and after obtaining the final sequence, we can transfer it to the shared database for annotation of this species.

[0106] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A data processing method for identifying DNA and nucleic acid organisms using biological nanopores, characterized by comprising the following steps: Step 1: Based on the characteristics of bio-nanopores, the DNA sequence to be detected is divided into combinations of several bases, and then the DNA sequence to be detected is detected by bio-nanopores; the raw current change data of the DNA sequence to be detected is obtained. Step two: Correct invalid values ​​in the original current change data using a correction algorithm; Step 3: Perform high-frequency noise filtering on the corrected current change data; Step 4: Divide the processed current change data into orifice current and blockage current, and filter out events that meet the criteria. The event refers to a blocking current event generated when DNA pores are present. The screening of events includes the following steps: Step 1: First, based on a broad threshold detection, select an event duration Time>1S for extraction, and a current range between 0.1 and 0.75 of the detected orifice current. Step 2: Initialize the local aperture current value based on the number of detected events and the detected aperture current. Step 3: Then, analyze and further filter suitable events based on each local open state and each event; The event is further filtered using the following steps: Step (1) involves detecting the before and after states of each numbered event in the event: event_before and event_after, which represent the open states before and after the event, respectively, with an initial value of NaN; Step (2): If event is not the first state, the median value of the open states before event is event_before; if event is not the last state, the median value of the state after event is event_after. Step (3): If the current event number is greater than 2 and data exists in the first two open states, try to fit the data of the previous state using the exponential decay function and update it to event_before; Step (4): If the current event number is less than the open state number and the current state has data, then try to fit the data of the next state using the exponential decay function and update it to event_after; Step (5), exception handling: if event_before and event_after deviate abnormally from the 3sigma value of the open-state current detected by event, they are also set to NaN; Step 5: Perform level search and fitting for the events that meet the search criteria; first, identify the level, determine the boundary of the current resistance level generated by different base combinations, and then remove false levels. Step 6: Establish a reference level database, select DNA samples with known sequences as the test set, perform nanopore sequencing and level extraction on them, compare the level data of the test set samples with the reference level database to obtain the annotated DNA sequences, compare the annotated DNA sequences with the known actual sequences of the test set samples, calculate the sequence similarity, identify the type and source of mismatched bases, evaluate the accuracy of the reference database, and optimize and improve the reference database based on the evaluation results; Step 7: Based on the Needleman-Wunsch and Smith-Waterman sequence alignment algorithms, the level data obtained from the sequencing of the DNA to be tested in Step 5 is compared with the level data in the reference level database. The most similar sequence is found, the DNA sequence of the DNA to be tested is inferred, the level of the newly measured DNA sequence is added to the reference level database, and the median of different sequence combinations is recalculated to update and improve the reference level database and improve its accuracy.

2. The data processing method for identifying DNA and nucleic acid organisms using bio-nanopores according to claim 1, characterized in that, In step one, the DNA sequence to be detected is sampled at a frequency of 500k using a biological nanopore.

3. The data processing method for identifying DNA and nucleic acid organisms using bio-nanopores according to claim 1, characterized in that, In step two, invalid values ​​are those that exceed a preset percentage of normal values, and the correction algorithm includes one of median filtering and weighted average algorithm.

4. The data processing method for identifying DNA and nucleic acid organisms using bio-nanopores according to claim 1, characterized in that, In step three, a Bayesian low-pass filter is used to filter high-frequency noise from the corrected original current change data; the sampling frequency in this project is 500KHz, the filter uses a fourth-order Bessel filter, and the cutoff frequency is 50KHz.

5. The data processing method for identifying DNA and nucleic acid organisms using bio-nanopores according to claim 1, characterized in that, The establishment of the reference level database in step six includes the following steps: Step S1: Select DNA samples with known sequences for nanopore sequencing to obtain raw current data; Step S2: Process and analyze the current data to extract reliable level information; Step S3: Repeat S1-S2 multiple times to accumulate level data and establish a reference level database. Step S4: Associate the level in the reference level database with its corresponding known DNA sequence to form a "level-sequence" mapping relationship.

Images