Gene analysis method and system based on time-of-flight mass spectrometry detection

Abstract

The application belongs to the technical field of gene analysis, and discloses a gene analysis method and system based on time-of-flight mass spectrum detection, which comprises the following steps: collecting the signal intensity time sequence generated by each ion pulse of the sample to be detected in a time-of-flight mass spectrum system, and constructing a time-of-flight data set; extracting the time distribution characteristics of the energy intensive area in each time signal unit, and calculating the energy concentration curve; constructing a continuously changing signal trend section by analyzing the energy change direction of adjacent time sections in the energy concentration curve; identifying the sensitive peak value section, extracting the peak value duration section and local energy turning point in the trajectory, and establishing a response mode characteristic vector; fusing the response mode characteristic vectors of all detection sites, constructing a time gene judgment model, and outputting the judgment result of the gene sample. The application avoids the misjudgment and missed detection phenomenon caused by signal fluctuation and noise, improves the accuracy and stability of detection, and enhances the anti-false negative capability of the signal.

Description

Technical Field

[0001] This invention relates to the field of gene analysis technology, and more specifically, to gene analysis methods and systems based on time-of-flight mass spectrometry. Background Technology

[0002] Existing gene detection technologies typically rely on gene amplification techniques (such as PCR), gene chips, and real-time quantitative PCR. Although these technologies offer high sensitivity and accuracy, false negatives still exist in complex biological samples, especially low-abundance samples or those with background noise. Time-of-flight mass spectrometry (TOF-MS), a high-throughput and precise analytical technique, has made significant progress in gene analysis in recent years. By analyzing the flight time of ions from the sample to the detector, it can provide highly accurate molecular information, particularly in the analysis of multi-component samples, offering more detailed mass spectrometry characteristic data.

[0003] However, in actual testing, gene response signals may exhibit low intensity or instability due to factors such as sample dilution, inconsistent ionization efficiency, or instrument noise. These low-intensity signals may fail to exceed the preset detection threshold, thus being misjudged as negative. Simultaneously, background noise or interference from other non-target components can also affect the accurate identification of signals. Particularly in time-of-flight mass spectrometry (TOF-MS), noise signals may mask true positive signals, leading to their incorrect interpretation as no signal. Furthermore, when signal fluctuations are large or the interval between ion pulses is short, the method of segmenting signals based on a fixed time window may incorrectly divide ion responses belonging to the same signal source into multiple parts during short ion pulse intervals, resulting in signal loss or misidentification, causing some true positive signals to fail to be captured. This missegmentation may miss weak positive signals, leading to false negative results. Therefore, a gene analysis method and system based on time-of-flight mass spectrometry is urgently needed to address these issues. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a gene analysis method and system based on time-of-flight mass spectrometry detection, comprising: The signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system of the sample to be tested is collected, and the signal intensity time series are combined in time sequence to construct a time-of-flight dataset. Based on the flight time dataset, extract the temporal distribution characteristics of the energy-dense region in each time signal unit and calculate the energy concentration curve; By analyzing the energy change direction of adjacent time intervals in the energy concentration curve, a continuously changing signal trend segment is constructed. Candidate peak segments are identified by analyzing the consistency between the rate of change of energy gradient and the degree of inflection in the peak region within the signal trend segment. By analyzing the correlation between the distribution location and probability density of turning points in the candidate peak segments, the candidate peak segments are screened to determine the sensitive peak segments. The response pattern feature vectors of the detection sites in the sensitive peak region are extracted and fused to construct a time-based gene determination model, and the determination results of the gene samples are output.

[0005] Preferably, the calculation of the energy concentration curve includes: Each time signal unit in the flight time dataset is divided into equally spaced segments according to a time window, and the local energy density is calculated in each segment. The energy density time series is obtained by using a sliding time window. Traverse the energy density time series and select continuous time segments where the local energy density is higher than the preset density threshold as high-energy intervals; Calculate the rate of energy change within the high-energy range and identify the location of local energy peaks where the rate of energy change changes from positive to negative; By using the local energy peak positions within the high-energy range as nodes, and fitting all nodes, an energy concentration curve is obtained.

[0006] Preferably, constructing the continuously changing signal trend segment includes: Using the energy concentration curve as input, the energy difference and differential symbol of adjacent points are calculated for each time point in the signal strength time series. The differential symbols are then combined in time sequence to obtain the differential symbol sequence. Based on the transition rate of the differential symbols in the differential symbol sequence, identify the stable trend segment of the signal intensity time series and calculate the average energy change rate of the stable trend segment. Based on the degree of difference in the average energy change rate between adjacent stable trend segments, the turning point between stable trend segments is established. A trend description vector is constructed based on the difference sign, duration, and average energy change rate of each stable trend segment; Extract multiple stable trend segments between adjacent turning points, and combine the trend description vectors according to the temporal order of the stable trend segments to obtain a trend description vector sequence; Perform a continuous smoothing operation on the trend description vector sequence and output the trend segment of the signal.

[0007] Preferably, the identification of candidate peak segments includes: Calculate the average energy of the signal trend segment and retain the signal trend segment whose average energy is greater than a preset energy threshold; Calculate the energy gradient for the signal trend segment to obtain the average, maximum, and minimum values ​​of the energy gradient. The difference between the maximum and minimum values ​​of the energy gradient is calculated to obtain the range of the extreme values ​​of the energy gradient. The ratio of the extreme span of the energy gradient to the average value of the energy gradient is calculated, and the result is used as the energy inflection intensity of the signal trend segment. Candidate peak segments with energy transition intensities greater than a preset intensity threshold are marked as candidate peak segments, and the time points corresponding to the maximum energy gradient values ​​in the candidate peak segments are marked as transition time points.

[0008] Preferably, determining the sensitive peak segment includes: Identify the start and end times of the candidate peak segment, and calculate the time intervals between the start and end times and the turning point, denoted as the enhancement duration and decay duration, respectively. The ratio of enhancement duration to decay duration is calculated, and the result is used as the peak shape stability coefficient. Histogram statistics are performed on the peak shape stability coefficients of different candidate peak segments to identify the peak shape stability coefficients in the maximum probability density interval, and the corresponding candidate peak segments are marked as sensitive peak segments.

[0009] Preferably, the extraction and fusion of response pattern feature vectors of detection sites in the sensitive peak region includes: Based on the time-of-flight mass spectrometry data, the mean energy, standard deviation of energy, mean energy gradient, duration of the segment, and peak stability coefficient of each detection site in the sensitive peak segment are extracted, and a response mode feature vector is constructed to generate a set of response mode feature vectors in the stable peak concentration. The similarity of response pattern feature vectors between any two sites is measured to obtain feature similarity; the similarity measure is obtained by calculating the Euclidean distance between response pattern feature vectors. Initialize an empty matrix, and then use different detection sites as horizontal and vertical input parameters respectively. Fill the empty matrix with the feature similarity between the corresponding response pattern feature vectors to obtain the similarity matrix. The similarity matrix is ​​summed column by column, and the sums of different column terms are normalized. The normalized result is used as the fusion weight of the detection sites in the column. The weighted response index is calculated based on the fusion weight of the detection sites, the average energy, the standard deviation of the energy, and the peak shape stability coefficient. The distribution characteristics of different detection sites on the weighted response index were statistically analyzed, and a positive response threshold was established.

[0010] Preferably, establishing the positive response threshold includes: Calculate the mean and standard deviation of the response index based on the weighted response index of different detection sites; The difference between the mean and standard deviation of the response index is calculated, and the result is used as the low intensity threshold. Detection points are screened based on a low intensity threshold, and those with a weighted response index less than the low intensity threshold are marked as low intensity points. The proportion of weak signal points is counted and recorded as the low-intensity signal area ratio. The baseline positive response threshold is calculated based on the low-intensity signal area ratio. The baseline positive response threshold is corrected based on time continuity.

[0011] Preferably, the construction of the time-based gene determination model includes: The weighted response index is re-screened using the positive response threshold after time continuity correction, and the weighted response indexes below the positive response threshold are retained. The corresponding sensitive peak segments are then included in the candidate weak signal set. Extend the time window for each sensitive peak segment in the candidate weak signal set to obtain the extended peak segment; Calculate the energy gain rate of each segment based on the average segment energy of the sensitive peak segment and the extended peak segment; When the energy gain rate is greater than the preset gain threshold, add gain marker points on the extended peak segment; For extended peak segments with gain markers, the weighted response index of the detection sites is recalculated and denoted as the extended response index. For each detection site, a set of consistency constraints is constructed by analyzing the consistency distribution structure of the extended response index in consecutive detection batches. By introducing gene determination rules into the consistency constraint set, a gene determination model is obtained.

[0012] Preferably, constructing the consistency constraint set includes: In continuous detection batches, the extended response index of each detection site is recorded and combined in time series to construct an index time series set; Perform a first-order difference operation on each exponential time series to extract the exponential change sign, and perform time series combination on the exponential change sign to obtain the sign time series. The directional consistency rate is obtained by measuring the directional consistency of symbolic time series. Calculate the standard deviation of the extended response exponent and the mean of the duration of the exponential time series, and then calculate the ratio of the standard deviation of the extended response exponent and the mean of the duration to obtain the coefficient of variation of the response intensity. The directional consistency rate and the coefficient of variation of response intensity are weighted and fused, and the result of the weighted fusion is used as the time stability score of the detection site. Detection sites with time stability scores greater than a preset score threshold are included in the time consistency constraint set.

[0013] This invention also provides a gene analysis system based on time-of-flight mass spectrometry detection, applied to the aforementioned gene analysis method based on time-of-flight mass spectrometry detection, comprising: The data acquisition module collects the signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system for the sample under test, and performs time-series combination of the signal intensity time series to construct a time-of-flight dataset; The data fitting module extracts the temporal distribution characteristics of the energy-dense region in each time signal unit based on the flight time dataset and calculates the energy concentration curve. The signal trend recognition module constructs a continuously changing signal trend segment by analyzing the energy change direction of adjacent time intervals in the energy concentration curve. Signal segment analysis module one identifies candidate peak segments by analyzing the consistency between the rate of change of energy gradient and the degree of turning point in the peak region within the signal trend segment. Signal segment analysis module two filters candidate peak segments and determines sensitive peak segments by analyzing the correlation between the distribution location and probability density of turning points in the candidate peak segments. The gene determination module extracts and fuses the response pattern feature vectors of the detection sites in the sensitive peak region to construct a time-based gene determination model and outputs the determination results of the gene samples.

[0014] The technical effects and advantages of the gene analysis method and system based on time-of-flight mass spectrometry detection of this invention are as follows: (1) By extracting the energy density time series from the flight time dataset, an energy concentration curve is constructed, and combined with sliding time window statistical analysis, the high-energy interval and stable trend segment are accurately divided. By performing dynamic feature decomposition on the signal, establishing signal trend segments, calculating the energy change rate and sensitive peak segments, the misjudgment and missed detection caused by signal fluctuations and noise can be effectively avoided, thus improving the accuracy and stability of detection. The signal's ability to resist false negatives is enhanced, especially in complex backgrounds where a high positive recognition rate can still be maintained.

[0015] (2) By constructing a gene determination model, the weighted response index is used to screen the detection sites, and the consistency constraint set is combined to track the weak signals in time, thereby achieving accurate identification and long-term stability determination of the signals; further, the response intensity of the detection sites is fused and corrected using multiple batches of data, avoiding misjudgment caused by noise changes in a single detection; by dynamically correcting the positive response threshold, the consistency of detection between batches is ensured, making the identification of weak signal samples more accurate, reducing the risk of misjudgment and missed detection, and improving the overall detection throughput and reliability. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the gene analysis method based on time-of-flight mass spectrometry detection according to the present invention.

[0017] Figure 2 This is a schematic diagram of the process for calculating the energy concentration curve in the gene analysis method based on time-of-flight mass spectrometry of the present invention.

[0018] Figure 3 This is a schematic diagram of the process for determining sensitive peak regions in the gene analysis method based on time-of-flight mass spectrometry of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] This application provides a gene analysis method and system based on time-of-flight mass spectrometry. By calculating the energy concentration curve, a continuously changing signal trend segment is constructed; sensitive peak regions are identified, and response mode feature vectors are established; the response mode feature vectors of all detection sites are fused to construct a time-based gene determination model, outputting the determination results of gene samples; this avoids misjudgments and missed detections caused by signal fluctuations and noise, improving the accuracy and stability of detection; and enhancing the signal's resistance to false negatives.

[0021] Example 1, please refer to Figure 1 , Figure 2 and Figure 3 In Embodiment 1 of the present invention, the gene analysis method based on time-of-flight mass spectrometry is implemented in detail through the following steps: The signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system of the sample to be tested is collected, and the signal intensity time series are combined in time series to construct a time-of-flight dataset; wherein, the signal intensity time series is the data corresponding to the time when the ion arrives at the detector under the action of the electric field and the response intensity, which is used to describe the flight characteristics of the ion swarm; Based on the time-of-flight dataset, the temporal distribution characteristics of the energy-dense region in each time signal unit are extracted, and the energy concentration curve is calculated. In Embodiment 1 of the present invention, the calculation of the energy concentration curve includes: Each time signal unit in the flight time dataset is divided into equally spaced segments according to a time window, and the local energy density is calculated in each segment. The energy density time series is obtained by using sliding time window statistics. The sliding time window statistics obtain the local energy density by setting a fixed-length time window and gradually shifting the calculation of the average intensity within the window. The energy density time series is traversed, and continuous time segments with local energy densities higher than a preset density threshold are selected as high-energy intervals. Specifically, by comparing the local energy density of each time segment with the preset density threshold, time segments with local energy densities greater than the preset threshold are marked as high-energy time segments. It is then determined whether a high-energy time segment is isolated (whether there are adjacent high-energy time segments). If a high-energy time segment is isolated, the marking is removed. If a high-energy time segment is not isolated, it is merged with adjacent high-energy time segments into a composite high-energy time segment, and the process continues to determine whether the composite high-energy time segment is isolated. If the composite high-energy time segment is not isolated, the merging of high-energy time segments continues until the composite high-energy time segment becomes isolated. The high-energy interval is generated based on the finally isolated composite high-energy time segment. The preset density threshold is set to 1.2 times the mean of the entire energy density time series. The time interval Δt and energy difference ΔE are calculated between adjacent high-energy intervals. When Δt is less than a preset time interval threshold τ (τ is 5% of the average pulse period) and ΔE is less than 3% of the total energy, the two intervals are merged into a continuous region, and the high-energy intervals are updated based on the continuous region. The time interval Δt and energy difference ΔE are used to determine whether adjacent high-energy intervals belong to the same energy release process, the preset time interval threshold τ is used to limit the time continuity condition, and the energy difference threshold is used to control the intensity smoothness. Through joint judgment, it is possible to prevent the signal from being incorrectly segmented due to small time breaks and improve the continuity of the energy curve. Calculate the rate of energy change within the high-energy range and identify the location of the local energy peak where the rate of energy change changes from positive to negative. The rate of energy change is obtained by calculating the first derivative of the energy density. The zero point represents the reversal point of the rate of energy change, that is, the time point when energy growth stops and begins to decay. The positions of local energy peaks within the high-energy range are used as nodes, and all nodes are fitted to obtain the energy concentration curve. The energy concentration curve is formed by the change in energy density along the time axis and is used to characterize the clustering of the signal in the time dimension, reflecting the degree of concentration of ion response energy in different time periods. The fitting can be achieved by cubic spline interpolation, which achieves a smooth connection of energy density through continuous curve fitting, while retaining the derivative information at each peak for determining the direction of subsequent signal trend segments, ensuring the continuity of the time dimension and numerical stability, thereby maintaining high anti-false negative performance under signal sparse conditions. By analyzing the energy change direction of adjacent time intervals in the energy concentration curve, a continuously changing signal trend segment is constructed; the signal trend segment is the result of the directional extension of the energy concentration curve in the time domain, which is used to describe the change pattern of ion response. In Embodiment 1 of the present invention, constructing a continuously changing signal trend segment includes: Using the energy concentration curve as input, the energy difference and differential symbol of adjacent points are calculated for each time point in the signal strength time series. The differential symbols are then combined in time sequence to obtain the differential symbol sequence. Among them, the difference symbol sequence is the first-order difference symbolization of the energy concentration curve, used to describe the direction of energy growth or decay in the time dimension; the symbol "+" indicates an upward trend and "-" indicates a downward trend; Based on the transition rate of differential symbols in the differential symbol sequence, the stable trend segment of the signal strength time series is identified, and the average energy change rate of the stable trend segment is calculated. Specifically, this involves counting the length of consecutive segments with the same symbol in the differential symbol sequence. When the length is greater than a preset length threshold (3% of the average pulse sampling length), the corresponding segment in the signal strength time series is defined as a stable trend segment. Here, a stable trend segment is a time interval in which the energy change direction remains consistent. The preset length threshold is used to filter out transient fluctuations caused by random noise. The average energy change rate is obtained by dividing the energy difference between the beginning and end of the stable trend segment by the duration. Based on the degree of difference in the average energy change rate between adjacent stable trend segments, a turning point between stable trend segments is established. Specifically, this involves calculating the absolute difference between the average energy change rates of adjacent stable trend segments. If the absolute difference exceeds a preset rate change threshold, a turning point is set at the boundary between the two segments. The preset change threshold is set to 1.5 times the standard deviation of the global average energy change rate, which is used to determine the true trend reversal. A trend description vector is constructed based on the difference sign, duration, and average energy change rate of each stable trend segment; Multiple stable trend segments between adjacent turning points are extracted, and the trend description vectors are combined according to the temporal order of the stable trend segments to obtain a trend description vector sequence; the trend description vector sequence describes and expresses the dynamic structure of energy evolution over time. A continuous smoothing operation is performed on the trend description vector sequence to output the signal trend segment. Specifically, when the difference signs in adjacent trend description vector segments are the same and the duration is less than half of a preset length threshold, their corresponding stable trend segments are merged, and the merged stable trend segment is taken as the signal trend segment. This is accomplished by checking for short-term repetitive changes in the same direction and merging segments to reduce redundant segments caused by sampling jitter. Candidate peak segments are identified by analyzing the consistency between the rate of change of energy gradient and the degree of inflection in the peak region within the signal trend segment. In Embodiment 1 of the present invention, identifying candidate peak segments includes: Calculate the average energy of the signal trend segments and retain signal trend segments with an average energy greater than a preset energy threshold; whereby, the formula is... The energy threshold was calculated. ; The global average energy; The global standard deviation is... An empirical factor between 1.0 and 1.3 is used to control the sensitivity of peak identification. The smaller the value of β, the more lenient the detection. When the value is higher, only the high-energy range is retained. The candidate peak segment is the time segment where the average energy is higher than the overall level. Calculate the energy gradient for the signal trend segment to obtain the average, maximum, and minimum values ​​of the energy gradient. The difference between the maximum and minimum values ​​of the energy gradient is calculated to obtain the range of the extreme values ​​of the energy gradient. The ratio of the extreme span of the energy gradient to the average value of the energy gradient is calculated, and the result is used as the energy inflection intensity of the signal trend segment. Candidate peak segments with energy transition intensities greater than a preset intensity threshold are marked as candidate peak segments, and the time points corresponding to the maximum energy gradient within these segments are marked as transition time points. The energy gradient, representing the derivative of energy over time within a segment, characterizes the rate of signal rise and fall. By statistically analyzing the probability density of energy transition intensities in historical signal trend segments, the probabilities of the top three energy transition intensity intervals are normalized to obtain weighted weights for these intervals. The medians of the top three energy transition intensity intervals are then weighted according to these weights, and the calculated results are used as the preset intensity threshold. When the energy transition intensity exceeds the preset threshold, it indicates a significant transition in signal intensity. The transition time point represents the precise location where the ion response changes from enhancement to decay, describing the release-decrease transition point of the flight signal. By analyzing the correlation between the distribution of turning points and probability density in candidate peak segments, candidate peak segments are screened to determine sensitive peak segments. In Embodiment 1 of the present invention, determining the sensitive peak region includes: Identify the start and end times of the candidate peak segment, and calculate the time intervals between the start and end times and the turning point, denoted as the enhancement duration and decay duration, respectively. The ratio of enhancement duration to decay duration is calculated, and the result is used as the peak shape stability coefficient. Histogram statistics were performed on the peak shape stability coefficients of different candidate peak segments to identify the peak shape stability coefficients in the maximum probability density range, and the corresponding candidate peak segments were marked as sensitive peak segments; among them, the peak shape symmetry coefficient was used to evaluate the maximum probability of energy change before and after the peak, representing a stable and highly repeatable ion flight signal; To address the common problems of low throughput and false negatives in existing gene detection methods, this paper extracts the energy density time series from the time-of-flight dataset, constructs an energy concentration curve, and combines it with sliding time window statistical analysis to accurately divide high-energy intervals and stable trend segments. By performing dynamic feature decomposition of the signal, establishing signal trend segments, and calculating the energy change rate and sensitive peak segments, it can effectively avoid misjudgments and missed detections caused by signal fluctuations and noise, thereby improving the accuracy and stability of detection. It also enhances the signal's resistance to false negatives, especially maintaining a high positive recognition rate even in complex backgrounds. Based on the response pattern feature vector set, information fusion is performed on the response patterns of the detection sites to construct a time-based gene determination model and output the determination results of gene samples. In Embodiment 1 of the present invention, the extraction and fusion of response pattern feature vectors of detection sites in the sensitive peak region includes: Based on the time-of-flight mass spectrometry data, the mean energy, standard deviation of energy, mean energy gradient, duration of the segment, and peak stability coefficient of each detection site in the sensitive peak segment are extracted, and a response mode feature vector is constructed to generate a set of response mode feature vectors in the stable peak concentration. The response mode feature vector is a set of energy structure, stability, and morphological characteristics of each site in the time-of-flight mass spectrometry data. The similarity of response pattern feature vectors of any two loci is measured to obtain feature similarity. The similarity measure is obtained by calculating the Euclidean distance between response pattern feature vectors. Feature similarity is used to measure the consistency of response patterns between different loci. The value range is 0 to 1. The higher the similarity, the closer the energy and morphological responses of the two loci are, and they may belong to the same gene segment or co-regulated region. Initialize an empty matrix, and then use different detection sites as horizontal and vertical input parameters respectively. Fill the empty matrix with the feature similarity between the corresponding response pattern feature vectors to obtain the similarity matrix. The similarity matrix is ​​summed column by column, and the sums of different column items are normalized. The normalized result is used as the fusion weight of the detection sites in the column. The fusion weight reflects the importance of each detection site in the population consistency. Sites with high feature similarity will receive higher weights, which means that their response in the sample is more reliable. This ensures that strong and stable signals dominate the results, thereby improving the overall positive recognition rate of weak signal samples and reducing false negatives caused by local noise. The weighted response index is calculated based on the fusion weights, average energy, energy standard deviation, and peak shape stability coefficient of the detection sites. Specifically, during the initial detection phase of time-of-flight mass spectrometry, the fusion weights, average energy, energy standard deviation, and peak shape stability coefficient of the detection sites are obtained. Different weight combinations are used for weighted calculation, and the standard deviation of the resulting weighted response index is calculated. The set of weights corresponding to the minimum standard deviation is used as the weighting weights for calculating the weighted response index. The weighted response index is a statistical measure of multidimensional signal fusion, considering energy intensity, peak shape stability, and energy fluctuation simultaneously after weighting each detection site with fusion weights. The distribution characteristics of different detection sites on the weighted response index were statistically analyzed to establish a positive response threshold. In Embodiment 1 of the present invention, establishing a positive response threshold includes: Calculate the mean and standard deviation of the response index based on the weighted response index of different detection sites; The difference between the mean and standard deviation of the response index is calculated, and the result is used as the low intensity threshold. Detection points are screened based on a low intensity threshold, and those with a weighted response index less than the low intensity threshold are marked as low intensity points. The proportion of weak signal points is statistically analyzed and recorded as the low-intensity signal area ratio. A baseline positive response threshold is then calculated based on this low-intensity signal area ratio. Specifically, this includes: using the formula... The noise ratio was calculated. P represents the proportion of the low-intensity signal region; (using the formula...) The baseline positive response threshold was calculated. ; and These are the mean and standard deviation of the response index, respectively; k is the dynamic adjustment coefficient, obtained through the formula... Dynamic value generation; dynamic adjustment of coefficient k and noise ratio Inversely proportional, when the noise level increases, the coefficient k automatically decreases, causing the threshold to shift appropriately, so that weak positive signals still have a chance to be included in the judgment range in samples with high noise, thereby reducing the risk of missed detection; The baseline positive response threshold is corrected based on time continuity; the correction term is determined according to the difference between the baseline positive response threshold of the previous test batch and the current test batch; the correction term is calculated using the following formula: In the formula, This is the smoothing coefficient (taken as 0.3–0.5). The baseline positive response threshold for the previous batch; the correction item Compared with the baseline positive response threshold Summing is performed to obtain the final positive response threshold. By correcting the baseline positive response threshold, abrupt changes in the threshold between adjacent test batches are avoided, thereby maintaining the stability of the results and preventing boundary samples from being misjudged due to fluctuations. In Embodiment 1 of this invention, a time-gene determination model is constructed, including: The weighted response indices are re-screened using a positive response threshold corrected for time continuity. Weighted response indices below the positive response threshold are retained, and the corresponding sensitive peak segments are included in the candidate weak signal set. Specifically, this includes: screening for... A set of candidate weak signals is constructed by using a weighted response index within a certain range. The proportional coefficient of 0.8 specifies the response range within 20% of the threshold, ensuring that the selected signals are both statistically significant and not completely confirmed as positive. This set is obtained through training with historical weighted response indices. The time window of each sensitive peak segment in the candidate weak signal set is extended to obtain the extended peak segment. The time length of the extended window is determined based on the average interval of ion pulses. Specifically, the first and second parts of the sensitive peak segment are extended respectively, and the time length of the extended window can be 10% of the average interval of ion pulses. The time window extension is achieved by adding a fixed-length compensation interval at both ends of the original signal segment, introducing more contextual data points, so that the subsequent feature extraction can cover the signal attenuation and rise boundary, and prevent incomplete energy statistics due to window clipping. The energy gain rate for each segment is calculated based on the average energy of the sensitive peak segment and the extended peak segment; the energy gain rate is calculated using the formula... ;in, This represents the average original energy value of the sensitive peak region. To extend the average energy of the peak region; where the energy gain rate is used to measure the signal enhancement effect brought about by time extension; When the energy gain rate is greater than the preset gain threshold, a gain marker point is added to the extended peak segment. The preset gain threshold is obtained by normalizing the probability density of the historical energy gain rate and normalizing the probability of the top three energy gain rate intervals to obtain the weighted weight of the energy gain rate intervals. The median of the three energy gain rate intervals is calculated based on the weighted weight. For extended peak segments with gain markers, the weighted response index of the detection sites is recalculated and denoted as the extended response index; specifically, this includes using the formula... In the formula, To expand the response index; The original weighted response index is determined by statistically analyzing the response pattern characteristics of known positive samples, negative control samples, and historical clinical test samples in historical time-of-flight mass spectrometry detection data, and by testing and analyzing energy intensity, peak stability, and energy fluctuation range. The weighting coefficients are obtained through training based on the balance between historical signals and new energy. For each detection site, a set of consistency constraints is constructed by analyzing the consistency distribution structure of the extended response index in consecutive detection batches. In Embodiment 1 of the present invention, the construction of a consistency constraint set includes: In continuous detection batches, the extended response index of each detection site is recorded and combined in time series to construct an index time series set. The index time series set contains the response intensity records of different detection sites in different detection batches. Through the historical accumulation of continuous batches, the continuous performance of weak signals at multiple time points can be analyzed, providing a real trend for subsequent modeling. For each exponential time series, a first-order difference operation is performed to extract the exponential change sign. The exponential change signs are then combined temporally to obtain the symbolic time series. The first-order difference represents the magnitude of the change in the site response between adjacent detection batches. By extracting the symbolic time series, the direction of the increase or decrease in response intensity can be determined, and a discretized expression of the time trend can be obtained. The directional consistency of a symbolic time series is measured to obtain the directional consistency rate. Specifically, this includes: counting the number of times the exponent change sign is the same in adjacent batches and the total number of batches; subtracting one from the total number of batches to obtain the total number of adjacent batches; and calculating the ratio of the number of times the exponent change sign is the same to the total number of adjacent batches. The result is used as the directional consistency rate. Calculate the standard deviation of the extended response index and the mean of the duration of the exponential time series, and then calculate the ratio of the standard deviation of the extended response index and the mean of the duration to obtain the coefficient of variation of the response intensity. The coefficient of variation of the response intensity is used to measure the dispersion of the response intensity of the consistent site in the time dimension. The smaller the coefficient of variation of the response intensity, the lower the volatility. The directional consistency rate and the coefficient of variation of response intensity are weighted and fused, and the result of the weighted fusion is used as the time stability score of the detection site. Specifically, in the initial detection stage of time-of-flight mass spectrometry, the historical directional consistency rate and the coefficient of variation of response intensity are calculated based on the historical index time series of the detection site. The historical directional consistency rate and the coefficient of variation of response intensity are weighted using different weight combinations. The standard deviation of the obtained time stability score is calculated, and the set of weights corresponding to the minimum standard deviation is used as the weight for calculating the time stability score. Detection sites with time stability scores greater than a preset score threshold are included in the time consistency constraint set. The preset score threshold is obtained by normalizing the probability density of historical time stability scores and normalizing the probability of the top three time stability score intervals. The weighted weight of the time stability score interval is obtained by weighting the median of the three time stability score intervals according to the weighted weight. Gene determination rules are introduced into the consistency constraint set to obtain a gene determination model. Specifically, the gene determination rules include: generating a standard reference vector for detection sites within the consistency constraint set based on time stability score, extended response index, and energy gain rate; obtaining the sample vector of the corresponding detection site in the gene sample to be detected; calculating the Euclidean distance between the standard reference vector and the sample vector, denoted as the determination distance; comparing this distance with a preset determination distance threshold; if the determination distance is greater than the threshold, a suspected negative result is output; if the determination distance is not greater than the threshold, the gene test is determined to be positive; the determination distance threshold is determined based on standardized data of industry positive detection probabilities. To address the false negative problem caused by noise interference in traditional gene detection methods, this paper constructs a gene determination model, uses a weighted response index to screen detection sites, and combines a consistency constraint set to track weak signals over time, thereby achieving accurate signal identification and long-term stability determination. Furthermore, it utilizes multiple batches of data to fuse and correct the response intensity of detection sites, avoiding misjudgments caused by noise variations in a single test. By dynamically adjusting the positive response threshold, it ensures consistency across batches, making the identification of weak signal samples more accurate, reducing the risk of misjudgment and missed detection, and improving the overall detection throughput and reliability.

[0022] Example 2: In Example 2 of the present invention, the gene analysis system based on time-of-flight mass spectrometry is implemented based on the following modules: The data acquisition module collects the signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system for the sample under test, and performs time-series combination of the signal intensity time series to construct a time-of-flight dataset; The data fitting module extracts the temporal distribution characteristics of the energy-dense region in each time signal unit based on the flight time dataset and calculates the energy concentration curve. The signal trend recognition module constructs a continuously changing signal trend segment by analyzing the energy change direction of adjacent time intervals in the energy concentration curve. Signal segment analysis module one identifies candidate peak segments by analyzing the consistency between the rate of change of energy gradient and the degree of turning point in the peak region within the signal trend segment. Signal segment analysis module two filters candidate peak segments and determines sensitive peak segments by analyzing the correlation between the distribution location and probability density of turning points in the candidate peak segments. The gene determination module extracts and fuses the response pattern feature vectors of the detection sites in the sensitive peak region to construct a time-based gene determination model and outputs the determination results of the gene samples.

[0023] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0024] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0025] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0026] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A gene analysis method based on time-of-flight mass spectrometry, characterized in that, include: The signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system of the sample to be tested is collected, and the signal intensity time series are combined in time sequence to construct a time-of-flight dataset. Based on the flight time dataset, extract the temporal distribution characteristics of the energy-dense region in each time signal unit and calculate the energy concentration curve; By analyzing the energy change direction of adjacent time intervals in the energy concentration curve, a continuously changing signal trend segment is constructed. Candidate peak segments are identified by analyzing the consistency between the rate of change of energy gradient and the degree of inflection in the peak region within the signal trend segment. By analyzing the correlation between the distribution location and probability density of turning points in the candidate peak segments, the candidate peak segments are screened to determine the sensitive peak segments. The response pattern feature vectors of the detection sites in the sensitive peak region are extracted and fused to construct a time-based gene determination model, and the determination results of the gene samples are output.

2. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 1, characterized in that, The calculation of the energy concentration curve includes: Each time signal unit in the flight time dataset is divided into equally spaced segments according to a time window, and the local energy density is calculated in each segment. The energy density time series is obtained by using a sliding time window. Traverse the energy density time series and select continuous time segments where the local energy density is higher than the preset density threshold as high-energy intervals; Calculate the rate of energy change within the high-energy range and identify the location of local energy peaks where the rate of energy change changes from positive to negative; By using the local energy peak positions within the high-energy range as nodes, and fitting all nodes, an energy concentration curve is obtained.

3. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 2, characterized in that, The construction of continuously changing signal trend segments includes: Using the energy concentration curve as input, the energy difference and differential symbol of adjacent points are calculated for each time point in the signal strength time series. The differential symbols are then combined in time sequence to obtain the differential symbol sequence. Based on the transition rate of the differential symbols in the differential symbol sequence, identify the stable trend segment of the signal intensity time series and calculate the average energy change rate of the stable trend segment. Based on the degree of difference in the average energy change rate between adjacent stable trend segments, the turning point between stable trend segments is established. A trend description vector is constructed based on the difference sign, duration, and average energy change rate of each stable trend segment; Extract multiple stable trend segments between adjacent turning points, and combine the trend description vectors according to the temporal order of the stable trend segments to obtain a trend description vector sequence; Perform a continuous smoothing operation on the trend description vector sequence and output the trend segment of the signal.

4. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 3, characterized in that, The identification of candidate peak segments includes: Calculate the average energy of the signal trend segment and retain the signal trend segment whose average energy is greater than a preset energy threshold; Calculate the energy gradient for the signal trend segment to obtain the average, maximum, and minimum values ​​of the energy gradient. The difference between the maximum and minimum values ​​of the energy gradient is calculated to obtain the range of the extreme values ​​of the energy gradient. The ratio of the extreme span of the energy gradient to the average value of the energy gradient is calculated, and the result is used as the energy inflection intensity of the signal trend segment. Candidate peak segments with energy transition intensities greater than a preset intensity threshold are marked as candidate peak segments, and the time points corresponding to the maximum energy gradient values ​​in the candidate peak segments are marked as transition time points.

5. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 4, characterized in that, The determination of the sensitive peak segment includes: Identify the start and end times of the candidate peak segment, and calculate the time intervals between the start and end times and the turning point, denoted as the enhancement duration and decay duration, respectively. The ratio of enhancement duration to decay duration is calculated, and the result is used as the peak shape stability coefficient. Histogram statistics are performed on the peak shape stability coefficients of different candidate peak segments to identify the peak shape stability coefficients in the maximum probability density interval, and the corresponding candidate peak segments are marked as sensitive peak segments.

6. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 5, characterized in that, The extraction and fusion of response pattern feature vectors of detection sites in sensitive peak regions includes: Based on the time-of-flight mass spectrometry data, the mean energy, standard deviation of energy, mean energy gradient, duration of the segment, and peak stability coefficient of each detection site in the sensitive peak segment are extracted, and a response mode feature vector is constructed to generate a set of response mode feature vectors in the stable peak concentration. The similarity of response pattern feature vectors between any two sites is measured to obtain feature similarity; the similarity measure is obtained by calculating the Euclidean distance between response pattern feature vectors. Initialize an empty matrix, and then use different detection sites as horizontal and vertical input parameters respectively. Fill the empty matrix with the feature similarity between the corresponding response pattern feature vectors to obtain the similarity matrix. The similarity matrix is ​​summed column by column, and the sums of different column terms are normalized. The normalized result is used as the fusion weight of the detection sites in the column. The weighted response index is calculated based on the fusion weight of the detection sites, the average energy, the standard deviation of the energy, and the peak shape stability coefficient. The distribution characteristics of different detection sites on the weighted response index were statistically analyzed, and a positive response threshold was established.

7. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 6, characterized in that, The establishment of the positive response threshold includes: Calculate the mean and standard deviation of the response index based on the weighted response index of different detection sites; The difference between the mean and standard deviation of the response index is calculated, and the result is used as the low intensity threshold. Detection points are screened based on a low intensity threshold, and those with a weighted response index less than the low intensity threshold are marked as low intensity points. The proportion of weak signal points is counted and recorded as the low-intensity signal area ratio. The baseline positive response threshold is calculated based on the low-intensity signal area ratio. The baseline positive response threshold is corrected based on time continuity.

8. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 7, characterized in that, The constructed time-gene determination model includes: The weighted response index is re-screened using the positive response threshold after time continuity correction, and the weighted response indexes below the positive response threshold are retained. The corresponding sensitive peak segments are then included in the candidate weak signal set. Extend the time window for each sensitive peak segment in the candidate weak signal set to obtain the extended peak segment; Calculate the energy gain rate of each segment based on the average segment energy of the sensitive peak segment and the extended peak segment; When the energy gain rate is greater than the preset gain threshold, add gain marker points on the extended peak segment; For extended peak segments with gain markers, the weighted response index of the detection sites is recalculated and denoted as the extended response index. For each detection site, a set of consistency constraints is constructed by analyzing the consistency distribution structure of the extended response index in consecutive detection batches. By introducing gene determination rules into the consistency constraint set, a gene determination model is obtained.

9. The gene analysis method based on time-of-flight mass spectrometry detection according to claim 8, characterized in that, The construction of the consistency constraint set includes: In continuous detection batches, the extended response index of each detection site is recorded and combined in time series to construct an index time series set; Perform a first-order difference operation on each exponential time series to extract the exponential change sign, and perform time series combination on the exponential change sign to obtain the sign time series. The directional consistency rate is obtained by measuring the directional consistency of symbolic time series. Calculate the standard deviation of the extended response exponent and the mean of the duration of the exponential time series, and then calculate the ratio of the standard deviation of the extended response exponent and the mean of the duration to obtain the coefficient of variation of the response intensity. The directional consistency rate and the coefficient of variation of response intensity are weighted and fused, and the result of the weighted fusion is used as the time stability score of the detection site. Detection sites with time stability scores greater than a preset score threshold are included in the time consistency constraint set.

10. A gene analysis system based on time-of-flight mass spectrometry, applied to the gene analysis method based on time-of-flight mass spectrometry as described in any one of claims 1 to 9, characterized in that, The data acquisition module collects the signal intensity time series generated by each ion pulse in the time-of-flight mass spectrometry system for the sample under test, and performs time-series combination of the signal intensity time series to construct a time-of-flight dataset; The data fitting module extracts the temporal distribution characteristics of the energy-dense region in each time signal unit based on the flight time dataset and calculates the energy concentration curve. The signal trend recognition module constructs a continuously changing signal trend segment by analyzing the energy change direction of adjacent time intervals in the energy concentration curve. Signal segment analysis module one identifies candidate peak segments by analyzing the consistency between the rate of change of energy gradient and the degree of turning point in the peak region within the signal trend segment. Signal segment analysis module two filters candidate peak segments and determines sensitive peak segments by analyzing the correlation between the distribution location and probability density of turning points in the candidate peak segments. The gene determination module extracts and fuses the response pattern feature vectors of the detection sites in the sensitive peak region to construct a time-based gene determination model and outputs the determination results of the gene samples.

Images