Intelligent control method and system of photoelectric synergistic enzymatic gene synthesis chip

Abstract

This invention belongs to the field of intelligent control technology for gene synthesis chips, and provides an intelligent control method and system for photoelectric synergistic enzymatic gene synthesis chips. The method involves applying a controllable light signal to the target site to deprotect the protecting group; switching the fluorescence sensing unit to a quenched state during the light signal application to block excitation light scattering interference; delivering fluorescently labeled base monomers for base extension under the catalysis of immobilized enzymes; injecting a washing solution to remove residues; after confirming that the light signal is off and the system is in a steady state, switching the fluorescence sensing unit to a Geiger state to collect fluorescence signals and identify four base types; applying a photoexcision signal; and using the characteristic parameters of the acquired fluorescence signals to evaluate the synthesis quality using an inference model, identifying the four synthesis working states, and then performing closed-loop error correction. This invention enables on-demand gene synthesis, base-by-base in-situ verification, and adaptive error correction.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for gene synthesis chips, specifically relating to an intelligent control method and system for photoelectric synergistic enzymatic gene synthesis chips. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Enzymatic DNA Synthesis (EDS) utilizes terminal deoxynucleotidyl transferase (TdT) to directly add bases to the ends of primers. Compared with the traditional phosphoramidite chemical synthesis method, it has the advantages of mild reaction conditions, no need for large amounts of organic solvents, and the ability to synthesize longer fragments. It has become an important technical route for chip-based gene synthesis platforms.

[0004] Current solutions integrate photocontrolled deprotection mechanisms with enzymatic extension into an array chip. By applying a controllable light signal, the protecting groups at the primer ends undergo a deprotection reaction, and then fluorescently labeled base monomers are delivered to complete the extension.

[0005] However, existing chip-level enzymatic gene synthesis control schemes still have the following problems: First, during the application of light signals, excitation light scattering and the afterglow after the light pulse is turned off can interfere with the fluorescence sensing unit located below the reaction site. Existing schemes lack a mechanism to synchronously switch the state of the fluorescence sensing unit during the application of light signals to eliminate the above interference. Second, when the same light source array needs to undertake two different photochemical reaction tasks, namely photocontrolled deprotection and photoexcision of fluorescent groups, there is a risk of overlapping light pulse doses required for the two working modes. Existing schemes lack an effective dose isolation mechanism to prevent mode crosstalk. Third, existing schemes mostly adopt offline detection methods after synthesis is completed, lacking a complete control scheme that performs in-situ fluorescence detection in each base cycle and combines intelligent inference to achieve real-time quality assessment and adaptive closed-loop error correction. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes an intelligent control method and system for photoelectric synergistic enzymatic gene synthesis chips. This invention enables in-situ real-time detection and intelligent closed-loop error correction, photoelectric deprotection and fluorescence detection timing interference, and improves the control accuracy of multi-site parallel synthesis.

[0007] According to some embodiments, the present invention adopts the following technical solution: A smart control method for a photoelectric co-catalyzed enzymatic gene synthesis chip, applied to at least one base cycle, includes the following steps: S1. A controllable light signal is applied to the target site of the photoelectric co-enzyme-catalyzed gene synthesis chip with a deprotection light pulse dose, so that the protecting group at the end of the primer fixed at the target site undergoes a deprotection reaction, exposing the active reactive group. S2. During the application of the controllable light signal, the fluorescence sensing unit below the target site is switched to a quenched state to block the interference of excitation light scattering on fluorescence detection. S3. Deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and connect the fluorescently labeled base monomers to the primer end under the catalysis of immobilized enzyme to complete the base extension; S4. Inject cleaning solution into the target site to remove unreacted base monomers, byproducts and residual enzyme reagents; S5. After confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to the Geiger state to collect the fluorescence signal of the extended base and identify the type of base synthesized at present based on the characteristic fluorescence wavelength corresponding to different base types. S6. Apply a photoexcision signal to the target site with a fluorescent group excision light pulse dose to cause the enzymatically cleavable linker arm covalently attached to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; S7. Based on the characteristic parameters of the fluorescence signal, use a pre-trained inference model to evaluate the synthesis quality of the current base cycle; S8. If the synthesis quality is normal, proceed to the next base cycle. If the synthesis quality is abnormal, adjust the optical signal parameters and re-execute the local deprotection reaction on the target site. If the synthesis quality fails and cannot be recovered, mark the target site as an abandoned site and skip the subsequent cycle.

[0008] As an alternative implementation, the deprotection pulse dose belongs to a first dose region, the fluorescent group excision pulse dose belongs to a second dose region, and a dose isolation zone is provided between the first dose region and the second dose region.

[0009] As a further defined embodiment, the relative value range of the light pulse dose in the second dose region is [1, 10], the relative value range of the light pulse dose in the first dose region is [20, 100], and the range of the dose isolation band is (10, 20).

[0010] As an alternative implementation, the controllable optical signal is ultraviolet light with a wavelength range of 350nm-380nm, an adjustable optical pulse width of 1ms-100ms, and a multi-level quantization of light intensity.

[0011] As an alternative implementation, the process of applying a controllable light signal to the target site of the reaction region of the photoelectro-co-enzymatic gene synthesis chip with a deprotected light pulse dose includes employing at least one addressing mode among single-point independent addressing, row batch addressing, column batch addressing, or global addressing, wherein the addressing mode is determined based on the site distribution characteristics of the current synthesis task.

[0012] As an alternative implementation, the fluorescence sensing unit adopts a 2×2 four-quadrant array structure, with each of the four quadrants configured with a characteristic fluorescence wavelength detection channel corresponding to one of the four base types.

[0013] As a further defined implementation, the relative response intensity is obtained by normalizing the photon count rate in the four quadrants within the detection window, and the type of base synthesized at the moment is identified based on the relative response intensity in each quadrant.

[0014] As a further defined implementation, it also includes recording the arrival time of the fluorescence signal photons using time-correlated single-photon counting technology, constructing a fluorescence lifetime histogram, and combining the characteristics of the two dimensions of fluorescence relative response intensity and fluorescence lifetime to distinguish four synthesis states: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination.

[0015] As a further defined implementation method, the specific process for distinguishing four synthesis states—correct synthesis, no synthesis, incorrect synthesis, and cross-contamination—by combining the characteristics of both relative fluorescence response intensity and fluorescence lifetime includes: If the relative response intensity of a certain quadrant exceeds the preset threshold, and the response intensity of the other three quadrants is lower than the cross-interference threshold, it is determined that the current synthesized base type corresponds to the base in that quadrant. If the response intensity in two or more quadrants exceeds the threshold, it is determined to be cross-contamination; If the response intensity in all quadrants is below the minimum detection threshold or there is no fluorescence signal, it is determined that it has not been synthesized; If the deviation between the fluorescence lifetime and the standard lifetime of the dye at the primer end is less than the set threshold, it is considered a correct synthesis; If the fluorescence lifetime deviates from the standard lifetime of the dye at the primer end by more than the set range, it is judged as an incorrect synthesis; If the fluorescence lifetime histogram shows a bimodal or multimodal distribution, it is considered cross-contamination.

[0016] As an alternative implementation, the characteristic parameters of the fluorescence signal include at least one of fluorescence intensity, photon count rate, and fluorescence lifetime; The pre-trained inference model is a lightweight model, employing at least one of a decision tree model, a neural network model, or a classifier based on a preset threshold.

[0017] As an alternative implementation, the process of adjusting the optical signal parameters and re-performing the local deprotection reaction on the target site includes: adjusting the optical signal parameters by increasing the optical pulse width, increasing the optical intensity level, and changing to a double pulse sequence or a continuous pulse sequence, and setting an upper limit on the number of times the local deprotection reaction can be re-performed. If the synthesis quality is still abnormal after re-performing the reaction beyond the upper limit, the synthesis is determined to be a failure, the target site is marked as an abandoned site, and a washing solution is injected into the target site in the subsequent base cycle to remove unreacted base monomers, byproducts, and residual enzyme reagents.

[0018] As an alternative implementation, the method further includes the following steps: generating a global synthesis quality report after each base cycle; collecting products from reaction sites determined to be normal after all base cycles are completed; discharging waste liquid from reaction sites determined to be waste; and ensuring that normal products and waste products are completely separated in the physical channel.

[0019] An intelligent control system for a photoelectric co-catalyzed enzymatic gene synthesis chip includes: The excitation light source driving module is configured to apply a controllable light signal to the target site of the reaction region of the photoelectro-co-enzymatic gene synthesis chip at a deprotection light pulse dose, causing the protecting group at the end of the primer fixed at the target site to undergo a deprotection reaction, exposing the active reactive group; and to apply a photocleavage signal to the target site at a fluorescent group cleavage light pulse dose, causing the enzymatically cleavable linker covalently linked to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; The fluorescence sensing control module is configured to switch the fluorescence sensing unit below the target site to a quenched state during the application of the controllable light signal to block the interference of excitation light scattering on fluorescence detection; after confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to a Geiger state to collect the fluorescence signal of the extended base. The base delivery module is configured to deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and to connect the fluorescently labeled base monomers to the primer ends under the catalysis of immobilized enzymes to complete base extension; The cleaning control module is configured to inject cleaning solution into the target site to remove unreacted base monomers, byproducts, and residual enzyme reagents. The AI ​​inference module is configured to identify the type of base being synthesized based on the characteristic fluorescence wavelengths corresponding to different base types; based on the characteristic parameters of the fluorescence signal, it uses a pre-trained inference model to evaluate the synthesis quality of the current base cycle; if the synthesis quality is normal, it proceeds to the next base cycle; if the synthesis quality is abnormal, it adjusts the optical signal parameters and re-executes the local deprotection reaction on the target site; if the synthesis quality fails and cannot be recovered, it marks the target site as an abandoned site and skips subsequent cycles. The timing scheduling module, electrically connected to other modules, is configured to uniformly schedule the input timing of the excitation light source driving module, fluorescence sensing control module, base delivery module, cleaning control module, and AI inference module, achieving site-level independent control and global parallel collaboration.

[0020] As an alternative implementation, a dose isolation band management module is also included, configured to limit the deprotection pulse dose to a first dose region and the phosphor removal pulse dose to a second dose region, wherein a dose isolation band is provided between the first and second dose regions, wherein the relative value range of the pulse dose in the second dose region is [1, 10], the relative value range of the pulse dose in the first dose region is [20, 100], and the range of the dose isolation band is (10, 20).

[0021] As an alternative implementation, the AI ​​inference module is configured to calculate the photon count rate in the four quadrants within the detection window, normalize it to obtain the relative response intensity, and identify the type of base synthesized at the moment based on the relative response intensity of each quadrant. The arrival time of the fluorescence signal photons was recorded by time-correlated single-photon counting technology, and a fluorescence lifetime histogram was constructed. By combining the characteristics of the two dimensions of fluorescence relative response intensity and fluorescence lifetime, four synthesis states were distinguished: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination.

[0022] As a further defined implementation, the AI ​​inference module is configured to: if the relative response intensity of a certain quadrant exceeds a preset threshold, and the response intensity of the other three quadrants is lower than the cross-interference threshold, determine that the base type of the currently synthesized base corresponds to the base in that quadrant. If the response intensity in two or more quadrants exceeds the threshold, or if the fluorescence lifetime histogram shows a bimodal or multimodal distribution, it is determined to be cross-contamination. If the response intensity in all quadrants is below the minimum detection threshold or there is no fluorescence signal, it is determined that it has not been synthesized; If the deviation between the fluorescence lifetime and the standard lifetime of the dye at the primer end is less than the set threshold, it is considered a correct synthesis; If the fluorescence lifetime deviates from the standard lifetime of the dye at the primer end by more than the set range, it is judged as an incorrect synthesis.

[0023] As an alternative implementation, the AI ​​inference model is deployed on the edge computing unit of the optoelectronic co-enzymatic gene synthesis chip.

[0024] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention achieves base-by-base in-situ verification and real-time error correction in the synthesis process by performing in-situ fluorescence detection and base type identification within each base cycle, and by evaluating the synthesis quality in real time and triggering local retry or discard markers. This avoids the problem of errors being passed downstream in existing offline detection schemes and significantly improves the fidelity of long fragment synthesis.

[0025] This invention effectively blocks the interference of excitation light scattering and afterglow on the single-photon detector by using a synchronous quenching mechanism that shares the same clock source with the quenching signal of the optical pulse signal and the quenching signal of the fluorescence sensing unit, and a timing design that switches to the Geiger state after the optical signal is turned off, through cleaning and waiting for the afterglow decay period. This ensures that the signal-to-noise ratio of the fluorescence signal meets the requirements for base recognition and solves the problem of lacking an excitation light interference elimination mechanism in the prior art.

[0026] This invention achieves the sharing of the same target site through photoexcision of the fluorescent group and photocontrolled deprotection. A dose isolation zone management module is set between the second dose zone [1,10] and the first dose zone [20,100] to ensure that a certain safety margin is maintained.

[0027] This invention deploys a lightweight AI inference model on the edge computing unit of a chip, and outputs a synthesis quality score and error correction suggestions in real time based on the three-dimensional feature parameter vectors of normalized fluorescence intensity, photon count rate, and fluorescence lifetime. The model weights are updated periodically through an incremental learning mechanism to adapt to chip aging and reagent batch differences, thus realizing the leap from fixed configuration to adaptive optimization of synthesis parameters and improving the robustness of the system to process fluctuations.

[0028] This invention uses a timing scheduling module to uniformly schedule the optical signal parameters, base delivery timing, and fluorescence detection timing of each reaction site. It supports four modes: single-point independent addressing, row batch addressing, column batch addressing, and global addressing. This enables site-level independent control and global parallel collaboration, taking into account the flexible needs of fine local regulation and high-throughput batch operation, and supports multi-site parallel synthesis of arbitrary sequences.

[0029] This invention uses base monomer reagents with different fluorescent staining as extension substrates. The fluorescent labeling is completed by base extension, and no additional labeling reaction step is required during detection, providing a physical basis for the synchronous recognition of four-quadrant arrays.

[0030] This invention combines fluorescence intensity and fluorescence lifetime as two dimensions, and uses time-correlated single-photon counting technology to construct a fluorescence lifetime histogram, which can distinguish four synthesis states: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination. Compared with single fluorescence intensity judgment, it significantly reduces the false judgment rate, provides richer decision-making basis for AI inference models, and further improves the accuracy of synthesis quality assessment.

[0031] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0032] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0033] Figure 1 A flowchart illustrating an embodiment of an intelligent control method for a photoelectric co-catalyzed enzymatic gene synthesis chip; Figure 2 This is a schematic diagram illustrating four addressing modes in one embodiment; Figure 3 This is a time-series diagram showing the relationship between optically controlled deprotection-quenching synchronization and fluorescence detection in one embodiment. Figure 4 This is a schematic diagram illustrating the working principle of a 2×2 four-quadrant fluorescence detection array according to one embodiment. Figure 5 This is a branch diagram of the closed-loop control logic for photoelectric synergistic enzymatic gene synthesis AI in one embodiment. Figure 6 This is a full-process timing waveform diagram of a single base cycle in one embodiment. Detailed Implementation

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0035] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0036] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0037] Where there is no conflict, the embodiments and features described in this application may be combined with each other.

[0038] Example 1 As described in the background section, the existing technology has the following technical problems: the lack of a synchronous quenching elimination mechanism to prevent interference from excitation light scattering and afterglow to the fluorescence sensing unit during the application of the light signal; the lack of an effective dose isolation mechanism due to the risk of overlapping light pulse doses required for the two working modes when the same light source array undertakes the photochemical reaction tasks of photocontrolled deprotection and photoexcision of fluorescent groups; and the lack of a complete control scheme that uses offline detection after synthesis, lacks in-situ fluorescence detection in each base cycle, and combines intelligent inference to achieve real-time quality assessment and adaptive closed-loop error correction.

[0039] To address the aforementioned problems, this embodiment provides an intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip, applied to at least one base cycle, such as... Figure 1 As shown, it includes the following steps: Step S1: Apply a controllable light signal to the target site of the photoelectric co-enzyme gene synthesis chip with a deprotection light pulse dose, so that the protecting group at the end of the primer fixed at the target site undergoes a deprotection reaction, exposing the active reactive group; Step S2: During the application of the controllable light signal, the fluorescence sensing unit below the target site is switched to a quenched state to block the interference of excitation light scattering on fluorescence detection. Step S3: Deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and ligate the fluorescently labeled base monomers to the primer ends under the catalysis of immobilized enzyme to complete the base extension; Step S4: Inject cleaning solution into the target site to remove unreacted base monomers, byproducts and residual enzyme reagents; Step S5: After confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to the Geiger state to collect the fluorescence signal of the extended base and identify the type of base synthesized at present based on the characteristic fluorescence wavelength corresponding to different base types. Step S6: Apply a photoexcision signal to the target site with a fluorescent group excision pulse dose to cause the enzymatically cleavable linker covalently attached to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; Step S7: Based on the characteristic parameters of the fluorescence signal, use a pre-trained inference model to evaluate the synthesis quality of the current base cycle; Step S8: If the synthesis quality is normal, proceed to the next base cycle; if the synthesis quality is abnormal, adjust the optical signal parameters and re-execute the local deprotection reaction on the target site; if the synthesis quality fails and cannot be recovered, mark the target site as an abandoned site and skip subsequent cycles.

[0040] Each step is described in detail below.

[0041] The first step is S1, in which a controllable light signal is applied to the target site in the chip reaction region, causing the primer ends fixed at the target site to undergo a deprotection reaction, exposing the active reactive groups.

[0042] In this embodiment, the on-chip integrated micro-LED array is fabricated on a silicon substrate using gallium nitride epitaxy, with a center wavelength of 365 nm and a full width at half maximum (FWHM) of ±5 nm. Each micro-LED has a luminous aperture of 50 μm × 50 μm, the array size is 100 × 100, and the center-to-center spacing between adjacent micro-LEDs is 200 μm.

[0043] In this embodiment, each micro-LED in the micro-LED array corresponds to a specific point in the gene reaction pool. By applying a controllable light signal to the target micro-LED, the protecting group at the end of the primer fixed at the target site undergoes a deprotection reaction, exposing the active reactive group.

[0044] The optical pulse width of the controllable optical signal is precisely controlled by pulse width modulation, with an adjustable range of 1ms to 100ms. For standard deprotection responses, a pulse width of 50ms is preferred; for sites with insufficient deprotection efficiency, the pulse width can be adaptively adjusted to the upper limit of 100ms.

[0045] In this embodiment, the light intensity levels are quantized into at least 32 levels, which is achieved by adjusting the driving current of the micro-LEDs. The lowest level corresponds to the minimum visible deprotection efficiency under a 1ms pulse, and the highest level corresponds to the saturation deprotection efficiency under a 100ms pulse. The driving current of the corresponding light intensity level is positively correlated with its level.

[0046] In some embodiments, the optimal light intensity level can be dynamically selected based on the optical parameters that yielded the best historical deprotection efficiency at each point.

[0047] like Figure 2 As shown, this embodiment supports four addressing modes: The single-point independent addressing mode applies an optical signal independently to any single unit in the micro-LED array through row and column decoding circuits, which is suitable for local retry or fine control of specific sites.

[0048] The row batch addressing mode applies optical signals synchronously to all micro LEDs in the same row, which is suitable for batch operations that simultaneously deprotect multiple bases.

[0049] The column batch addressing mode applies optical signals synchronously to all micro LEDs in the same column, which is suitable for scheduling synthesis tasks organized by column.

[0050] The global addressing mode applies optical signals to all micro LED units one by one, which is suitable for deprotection of all sites or global reset operation during the initialization phase.

[0051] The addressing mode can be selected automatically based on the site distribution characteristics of the current synthesis task, or it can be pre-configured and specified.

[0052] For example, if the sites of the current synthesis task are all in the same row, the row batch addressing mode can be selected.

[0053] The four addressing modes take into account the flexible needs of both single-point fine-tuning and batch high-throughput operation, and support multi-site parallel synthesis of arbitrary sequences.

[0054] In step S2, the timing synchronization mechanism for quenching the fluorescence sensing unit and applying the optical signal, and in step S5, the timing control for switching the fluorescence sensing unit to the Geiger state, are combined with... Figure 3 Please provide an explanation.

[0055] In this embodiment, the fluorescence sensing unit employs a single-photon avalanche diode (SPAD) array, with each SPAD unit equipped with an independent quench-recharge circuit. When the quench signal is valid, the reverse bias voltage of the SPAD is pulled down below the breakdown voltage, placing the SPAD in a quenched state and preventing the generation of avalanche current. This blocks the multiplication process of photogenerated carriers, effectively eliminating the interference of excitation light scattering on fluorescence detection.

[0056] like Figure 3 As shown, the micro-LED driving signal and the single-photon avalanche diode quenching signal are derived from the same phase-locked loop clock source, with a clock frequency of no less than 100MHz, ensuring that their time alignment accuracy is no greater than 10ns. At time t0, the rising edge of the micro-LED driving signal triggers the output of a light pulse, and the single-photon avalanche diode quenching signal switches to the active state within t0±5ns, and the single-photon avalanche diode enters the quenching mode.

[0057] At time t0+T_pulse, the falling edge of the micro-LED driving signal ends the optical pulse output; the single-photon avalanche diode quenching signal remains valid until t0+T_pulse+T_quench, where T_quench is the quenching duration and T_pulse is the pulse signal width.

[0058] At time t0 + T_pulse + T_quench, the quenching signal of the single-photon avalanche diode is released, and the single-photon avalanche diode switches to Geiger mode, ready to receive the fluorescence signal of the extended bases. The waiting time covers the afterglow decay time of the light pulse, the stabilization time of the cleaning liquid surface, and the stabilization time of the biological system, typically ranging from 10 ms to 50 ms. This is sufficient to ensure the complete elimination of the afterglow of the micro-LED and the stability of the biological reaction system, thus ensuring the reliability of single-photon detection.

[0059] In step S3, the fluorescently labeled base monomer is delivered and its extension reaction is catalyzed by the immobilized enzyme.

[0060] In each base cycle, a single type of fluorescently labeled deoxynucleotide is selected and delivered based on the base type at the current site of the target sequence. Each of the four base types is covalently coupled with a characteristic fluorescent dye via an enzymatically cleavable linker arm, forming a detectable base monomer: adenine corresponds to FAM, thymine to TAMRA, cytosine to ROX, and guanine to Cy5. As one implementation method, the fluorescent dye and base formulations shown in Table 1 are used. Only the fluorescently labeled deoxynucleotide corresponding to a single base type is delivered in each cycle. The four fluorescently labeled deoxynucleotides are not present simultaneously in the same reaction system to avoid non-selective random synthesis by terminal deoxynucleotidyl transferase.

[0061] Table 1. Formulations of fluorescent dyes and bases with wavelength separation

[0062] The enzymatically cleavable linker remains stable under elongation reaction conditions, ensuring the simultaneous synthesis of the fluorescent dye and base to the primer terminus under TdT catalysis. The core cleavage group of the enzymatically cleavable linker is an o-nitrobenzyl derivative with an absorption peak at 350 nm-370 nm and a photolysis quantum yield of approximately 0.2-0.5. Under 365 nm illumination, the o-nitrobenzyl group undergoes intramolecular rearrangement and cleavage, releasing the fluorescent group and simultaneously exposing a free hydroxyl group at the 3' position of the deoxynucleoside monophosphate.

[0063] The bottom of the microreactor is pre-modified with gold nanoparticles, and the terminal deoxynucleotidyl transferase is covalently fixed to the surface of the particles to form an immobilized enzyme catalytic layer.

[0064] The loading of terminal deoxynucleotidyl transferase was 0.1 U / μL to 1 U / μL, preferably 0.5 U / μL. The extension reaction was carried out at a constant temperature of 37±0.1℃ for 1 to 8 seconds, preferably 5 seconds. The reaction buffer consisted of 50 mM Tris-HCl (pH 7.5), 10 mM MgCl2, and 1 mM DTT. The extension reaction time was adjusted within the range of 1 to 8 seconds based on historical extension efficiency data.

[0065] After extension, a base with a fluorescent dye is added to the 3' end of the primer. This fluorescent group serves as a detection probe for both the synthesis and sequencing strategies, allowing the primer to enter the fluorescence detection stage without additional cross-linking or labeling steps.

[0066] It should be noted that the immobilized enzyme is not limited to terminal deoxynucleotidyl transferase, but also includes its mutants, fusion proteins or other DNA polymerases with similar catalytic activity.

[0067] In step S5, the fluorescence signal acquisition and base type identification mechanism, and in step S7, the AI ​​inference model's processing of the fluorescence signal, are combined with... Figure 4 Please provide an explanation.

[0068] like Figure 4 As shown, a 2×2 four-quadrant single-photon avalanche diode detection array is configured below each microreactor. Different fluorescence filters are configured in each of the four quadrants. Quadrant I has a band of 515nm-525nm, corresponding to adenine (FAM); Quadrant II band 575nm-585nm corresponds to thymine (TAMRA); Quadrant III has a band of 600nm-620nm, corresponding to cytosine (ROX); Quadrant IV has a bandwidth of 665nm-675nm, corresponding to guanine (Cy5).

[0069] Each quadrant contains at least one single-photon avalanche diode unit with a photon detection efficiency of no less than 30% in the range of 450nm to 700nm. The four-quadrant array shares the same quench-recharge circuit to ensure synchronous switching of the four channels.

[0070] After fluorescence signal acquisition, the base identification logic is executed: the photon count rate in each of the four quadrants within the detection window is calculated; the photon count rate in each of the four quadrants is normalized, and the relative response intensity vector (R_A, R_T, R_C, R_G) is calculated; if the relative response intensity in one quadrant exceeds a preset threshold (e.g., 0.6) and the response intensity in the other three quadrants is lower than the cross-interference threshold (e.g., 0.2), then the currently synthesized base type is determined to correspond to the base in that quadrant; if the response intensity in two or more quadrants exceeds the threshold, it is determined to be cross-contamination; if the response intensity in all quadrants is lower than the minimum detection threshold or there is no fluorescence signal, it is determined to be unsynthesized.

[0071] In some implementations, fluorescence lifetime analysis of the fluorescence signal is also performed using time-correlated single-photon counting technology. A fluorescence lifetime histogram is constructed by recording the arrival time of each photon; the fluorescence lifetime τ reflects the microenvironmental state of the fluorescent dye. When synthesized correctly, τ is close to the standard lifetime of the dye at the primer end (e.g., for FAM, τ≈4.0ns at the DNA end). There is no fluorescence signal when it is not synthesized; If the τ value deviates from the calibration value by ±20% during incorrect synthesis, it is considered abnormal. When cross-contamination occurs, the histogram shows a bimodal or multimodal distribution, and the τ value is dispersed.

[0072] By combining fluorescence intensity and fluorescence lifetime, the inference model can distinguish four synthesis states: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination, reducing the uncertainty of single-dimensional judgment.

[0073] In step S6, after completing the fourth stage fluorescence detection and AI evaluation, if it is determined that the reaction site needs to continue to the next round of synthesis, a fluorescent group photoexcision operation is performed. If it is determined to be abnormal or failed, the fluorescence excision is skipped, and the error correction or abandonment process is directly executed.

[0074] In this step, the same μLED array used for the controllable light signal in step S1 is used to apply a photoexcision signal to the target site with a fluorescent group excision light pulse dose, causing the enzymatically cleavable linker covalently attached to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group.

[0075] The dose of the fluorescent group-removed light pulse falls into the low-dose region (i.e., the second dose region), which is different from the deprotected light pulse dose that falls into the high-dose region (i.e., the first dose region) in step S1. The relative value of the light pulse dose in the low-dose region falls into the interval [1, 10], and the relative value of the light pulse dose in the high-dose region falls into the interval [20, 100]. A dose isolation band is set between the two, which is the prohibited output interval between the low-dose region and the high-dose region. The dose isolation band is the interval (10, 20).

[0076] In this embodiment, the light pulse dose is defined as the product of the light intensity level and the pulse width (normalized relative value). The light pulse width is adjustable from 1ms to 100ms. The light intensity level is divided into at least 32 quantization levels. The minimum light pulse dose can be the value of the minimum light intensity multiplied by the minimum pulse width, i.e., 1, and the maximum light pulse dose can be the value of the maximum light intensity multiplied by the maximum pulse width, i.e., 100, thus achieving normalization.

[0077] The micro-LED array integrated on the chip (center wavelength 365nm, HW50 ± 5nm) performs both photocontrolled deprotection and fluorescence excision. The photopulse dose is defined as the product of the light intensity level and the pulse width (normalized relative value). The fluorescence excision reaction requires a lower dose because the enzymatically cleavable linker is a single-molecule breakage reaction with a high photolysis quantum yield (Φ≈0.2-0.5), eliminating the need to drive a macroscopic deprotection reaction. The photocontrolled deprotection reaction requires a higher dose because (e.g., NPPOC) has a lower photolysis quantum yield (Φ≈0.005-0.01), requiring a higher photon flux to ensure sufficient exposure of the 3'-OH group.

[0078] To avoid mode crosstalk, the above-mentioned dose isolation band mechanism is used to divide the working dose into three intervals, and any combination of optical signals falling into the dose isolation band interval is prohibited from being output.

[0079] As a typical example, as shown in Table 2, the specific parameters can be designed as follows: Table 2 Optical Dosimetry Zoning Table

[0080] The basis for setting the dose isolation zone is: A safety margin of 2 times is maintained between the upper limit of the low dose region (10) and the lower limit of the high dose region (20). Even considering the ±10% temperature drift of the micro-LED drive current and clock jitter, mode out-of-range will not occur, fundamentally eliminating mode crosstalk between the two photochemical reactions.

[0081] In some embodiments, the cumulative dose at each reaction site can be recorded, where the cumulative dose D_accumulated(n) = , i=1 to n.

[0082] D_deprotect_i is the dose of the deprotection pulse in the i-th round, and D_cleave_i is the dose of the phosphor removal pulse in the i-th round.

[0083] When the cumulative dose at a certain site exceeds a preset threshold (e.g., preset threshold D_threshold=5000), the site is marked as a high-exposure site, and in subsequent cycles, low-dose region parameters are preferentially used for fluorescent excision; if the cumulative dose exceeds the safety limit (e.g., D_max=10000), the site is forcibly marked as an abandoned site to avoid DNA strand breakage or mutation due to cumulative light damage.

[0084] After photolysis and breakage, the 3' end of the primer regains its free hydroxyl group. The regeneration efficiency is confirmed by two methods: functional verification and background verification. Functional verification involves performing a cut-extension control cycle on the chip. If the terminal deoxynucleotidyl transferase can extend normally in the next round after cut (a new round of expected fluorescence signal is detected), it proves that the 3'-OH has been fully exposed. Background verification involves detecting the fluorescence background of the site after cut-out and washing. If the photon count rate is less than 5% of the minimum detection threshold, it is determined that the cut-out is complete.

[0085] In this embodiment, the core cleavage group of the enzymatically cleavable linker is an o-nitrobenzyl derivative with an absorption peak at 350 nm–370 nm and a photolysis quantum yield of approximately 0.2–0.5. Under 365 nm illumination, the o-nitrobenzyl group undergoes intramolecular rearrangement and cleavage, releasing a fluorophore and simultaneously exposing a free hydroxyl group (-OH) at the 3' position of the deoxyribonucleoside monophosphate.

[0086] In some embodiments, the enzymatically cleavable linker can be coumarin-4-ylmethyl ester or diarylethylene derivative, and its photolysis wavelength can be finely adjusted to 370nm–390nm to form a slight offset from the 365nm center wavelength of the deprotected light signal, further reducing the risk of crosstalk.

[0087] In step S7, the architecture of the AI ​​inference model, the input feature vector, the output quality score, and the model training and update mechanism are specifically as follows: Based on the computing power limitations of the chip's edge computing unit (e.g., using a RISC-V processor with a clock speed of 400MHz and equipped with on-chip memory), a lightweight architecture is selected for the AI ​​inference model.

[0088] In one implementation, a small neural network is used: the input layer has 3 nodes, corresponding to three feature parameters; the hidden layer has 2 layers, each with 8 to 16 nodes; the output layer has 4 nodes, corresponding to four synthesis states; the ReLU activation function and the Softmax output layer are used, and the number of parameters is controlled within 1000.

[0089] In another implementation, a threshold classifier based on statistical distribution is used: if the fluorescence intensity is greater than threshold 1 and the fluorescence lifetime is within the standard interval [τ_min, τ_max], it is determined to be a correct synthesis; if the fluorescence intensity is less than threshold 2, it is determined to be a non-synthesized synthesis.

[0090] The input feature parameter vector X for each reaction site is defined as X=[I_fluo, C_photon,τ_fluo], where I_fluo is the normalized fluorescence intensity, ranging from [0, 1], C_photon is the photon count rate (unit: counts / s), and τ_fluo is the fluorescence lifetime (unit: ns).

[0091] The AI ​​inference model outputs a synthesis quality score Q and error correction suggestions: Q∈[0, 1], Q≤0.8 is considered normal, 0.5≤Q<0.8 is considered abnormal, and Q<0.5 is considered a failure; error correction suggestions include increasing the light pulse width, increasing the light intensity level, using a double pulse sequence, or marking as discarded.

[0092] The AI ​​inference model is trained using supervised learning. The training dataset consists of fluorescence feature parameters from historical synthesis cycles and manually labeled synthesis states. The model periodically (e.g., every 1000 cycles) updates the weights through incremental learning to adapt to parameter drift caused by chip aging, reagent batch differences, etc., achieving a leap from fixed parameter configuration to adaptive parameter optimization.

[0093] In step S8, the closed-loop error correction strategy and management function are combined with Figure 5 Please provide an explanation.

[0094] When the AI ​​inference model determines that the synthesis at a certain reaction site is abnormal (e.g., 0.5 ≤ Q < 0.8), a local retry is performed: keeping the current sequence progress at that site unchanged, the pulse width is first increased from 50ms to 100ms; if the abnormality persists, the light intensity level is increased (e.g., from level 16 to level 24); if the abnormality still persists, a double-pulse sequence (two 50ms pulses, 10ms apart) is used instead of a single pulse. The maximum number of retries is 3; if the abnormality persists after 3 retries, the synthesis is considered a failure.

[0095] When a failure is detected (Q is less than 0.5 or the retry limit is exceeded), the coordinates of the site are entered into the discarded site register.

[0096] In all subsequent base cycles, this site skips steps S1 (photosensitive deprotection), S2 (quenching synchronization), and S3 (enzymatic base extension), and only performs step S4 (washing) to prevent defective products from continuing to extend and mixing into the final product. The sequence information of the discarded site is not included in the final product collection, but is retained in the log for subsequent analysis.

[0097] In addition, the fluorescence excision operation is executed immediately after the AI ​​inference model completes its inference, as shown in Table 3.

[0098] Table 3 Fluorescent group removal timing

[0099] Where t_AI_end is the time when the AI ​​inference of the inference model is completed, Δt_switch is the mode switching delay (<1ms), T_wash is the short cycle cleaning time (preferably 3s in this embodiment), and Δt_safe is the safety margin (10ms).

[0100] In this embodiment, after each base cycle, a global synthesis quality report is generated, which includes: statistics on the synthesis status of each reaction site (number and proportion of normal, abnormal, failed, and unsynthesized sites), error site coordinate mapping (row and column coordinates in a two-dimensional array), overall synthesis fidelity score (number of correct sites / total number of sites), and statistical distribution of average fluorescence intensity, photon count rate, and fluorescence lifetime. The global synthesis quality report is uploaded in real time, with a refresh frequency synchronized with the base cycle period.

[0101] After all base cycles are completed, for reaction sites determined to be synthesized normally, the product collection channel valve is opened, and the target DNA product is exported to the collection tube outside the chip by microfluidic pressure drive; for reaction sites determined to be discarded, the waste liquid discharge channel valve is opened, and the waste product is introduced into the waste liquid collection container; normal products and waste products are completely separated in the physical channel to avoid cross-contamination.

[0102] The above method can be applied to scenarios where gene synthesis and detection are performed using photoelectric synergistic enzymatic gene synthesis chips. To ensure a clearer solution, this embodiment may also use a photoelectric synergistic enzymatic gene synthesis chip structure, but this does not mean that the control method provided in this embodiment is only applicable to the above chip structure.

[0103] A photoelectric co-produced enzyme gene synthesis chip includes a light-controlled chip, which includes an addressing module, a μLED driving module, and a μLED array. The addressing module communicates with the servo chip and is used to receive addressing commands. The μLED driving module sends a driving signal to cause the μLED in the μLED array corresponding to the target site to emit excitation light to trigger the removal of the protecting group in the corresponding microreactor. The gene synthesis chip comprises a microfluidic channel layer, a microreactor array layer, and a product collection-waste treatment layer. The microreactor array layer contains multiple microreactors arranged in an array, each containing a terminal deoxynucleotidyl transferase or its engineered variant, configured to perform enzymatic base extension after photocontrolled deprotection. The microfluidic channel layer, located above the microreactor array layer, includes several reagent supply channels, enzyme channels, rinsing solution channels, and bubble removal channels to provide reagents, enzymes, washing solutions, and buffer solutions to each microreactor. The product collection-waste treatment layer, located below the microreactor array layer, includes waste collection channels and product collection cells, used to receive waste liquid and reaction products from each microreactor, respectively.

[0104] The fluorescence sensing control chip includes a SPAD array and a Geiger-quenching control circuit. The SPAD array is configured to perform single-photon detection of the fluorescence signal in the microreactor after the μLED array is turned off. The Geiger-quenching control circuit controls the corresponding SPAD detection module to be in a quenched state during μLED illumination. After the μLED is turned off, the Geiger-quenching circuit module controls the corresponding detection module to switch to Geiger mode for single-photon fluorescence detection. In this embodiment, a 2×2 four-quadrant SPAD array is configured below each microreactor. A fluorescent filter lens layer is set above the array to focus the fluorescence emitted by the microreactor onto the photodetector photosensitive area and filter out excitation light scattering and ambient stray light. A data processing layer is set below the SPAD array to record the arrival timestamp of each photon event, perform statistical histogram analysis on the photon events detected by each microreactor during the synthesis cycle, and extract fluorescence intensity, photon count rate, and fluorescence lifetime. Based on the extracted data, the synthesis quality of each microreactor is evaluated in real time using a preset threshold or a lightweight machine learning model. The evaluation results are sent to the servo chip via a communication interface. The servo chip is configured to control the temporal coordination of the photocontrol chip, the gene synthesis chip, and the fluorescence sensing control chip to achieve control of photocontrol deprotection-quenching synchronization-enzymatic extension-rinsing of residues-fluorescence detection-feedback error correction-fluorescent group removal; The light control chip, gene synthesis chip, fluorescence sensing control chip, and servo chip are 3D vertically stacked and packaged through TSV through-silicon vias and μBump microbump interconnect structures, and the μLED array, microreactor, and SPAD array are aligned in the vertical direction.

[0105] The aforementioned chiplet refers to a silicon die with a specific set of functions and a standardized die-to-die interface.

[0106] The aforementioned chip is mounted on a substrate.

[0107] In some embodiments, the servo chip includes a light-controlled chip servo module, a gene synthesis chip servo module, a fluorescence sensing and control chip servo module, a power management module, and an electrical signal interface module, wherein: The photocontrol particle servo module is communicatively connected to the photocontrol particle and is used to send corresponding addressing instructions and light pulse control instructions to the photocontrol particle according to the base type and target sequence information of the current synthesis cycle. After receiving the synthesis error information fed back by the fluorescence sensing photocontrol particle, it dynamically adjusts the photocontrol parameters at the error point. The gene synthesis core servo module communicates with an external pump valve via an external electrical signal interface, and is used to send control commands for reagent injection, rinsing solution injection, waste liquid discharge and product collection in a time sequence. The fluorescence sensing and control chip servo module is communicatively connected to the fluorescence sensing and control chip and is equipped with a quenching synchronization circuit, which is used to complete the global quenching of the photodetector within a predetermined time after the rising edge of the light source driving signal is triggered, and to configure the readout mode of the photodetector array during the fluorescence detection stage. The power management module includes a multi-channel voltage regulator and a DC-DC converter, which are used to provide stable power supplies of multiple voltage levels to each chip and control the power-on and power-off of each chip according to a safe timing sequence. The electrical signal interface module is used to communicate with an external host computer, report the synthesis status in real time, and receive gene sequence data and timing control commands.

[0108] Example 2 This embodiment uses a single base cycle as an example to demonstrate the complete temporal relationship and duration of steps S1 to S8 in the method provided in Embodiment 1, combined with... Figure 6 The timing parameters are set as follows: the target base type is adenine, the micro-LED array size is 100×100, and global addressing mode is used.

[0109] like Figure 6 As shown, the timing relationship of each stage is as follows: at t=0ms, step S1 starts, the micro LED outputs a deprotection pulse (light intensity level 20, pulse width 50ms), step S2 is synchronously effective, and the single-photon avalanche diode quenching signal switches to the effective state within t0±5ns; at t=50ms, deprotection ends, the micro LED turns off, and the quenching signal remains effective; from t=50ms to 5s is step S3, the enzymatic extension stage (delivery of FAM-labeled adenine deoxynucleotide, extension reaction time 5s); from t=5.05s to 8.05s is step S4, the washing stage (washing time 3s).

[0110] From t=8.05s to 8.06s is the waiting period for the liquid surface to stabilize (10ms); from t=8.06s to 8.07s is the fluorescence detection window of step S5 (single-photon avalanche diode Geiger mode, detection time 10ms), during which the fluorescence signal of the extended bases is collected and the four-quadrant base recognition logic is executed; from t=8.07s to 8.17s is the AI ​​inference stage of step S7 (edge ​​computing unit operation, inference time 100ms), during which the synthesis quality score Q and error correction suggestions are output; at t=8.17s, the AI ​​inference is completed, and if the judgment is normal, the process proceeds to step S6.

[0111] From t=8.17s to 8.18s is step S6, the fluorescent group removal stage, during which the micro-LED outputs a fluorescent removal pulse (light intensity level 8, pulse width 10ms, relative dose 5, falling into the low dose region [1, 10]), and the single-photon avalanche diode remains in the quenched state; from t=8.18s to 8.68s is the short-cycle cleaning stage (removing detached fluorophores, cleaning time 500ms); at t=8.68s, the next base cycle begins, and the target base type is switched to thymine.

[0112] In this embodiment, the total time for a single base cycle is approximately 8.63 s / base, with the fluorescence excision step accounting for only about 0.12 s, having an impact of less than 2% on the synthesis throughput. The above timing parameters verify that steps S1 to S8 are strictly cascaded and coordinated in time, with no timing conflicts between steps. The unified scheduling of optical signal parameters, base delivery timing, fluorescence detection timing, and fluorescence excision timing enables site-level independent control and global parallel coordination.

[0113] The complete base cycle sequence is shown in Table 4.

[0114] Table 4. Complete base cycle timing (including fluorescent group removal)

[0115] Example 3 An intelligent control system (or intelligent controller) for a photoelectric co-catalyzed enzymatic gene synthesis chip includes: The excitation light source driving module is configured to apply a controllable light signal to the target site of the reaction region of the photoelectro-co-enzymatic gene synthesis chip at a deprotection light pulse dose, causing the protecting group at the end of the primer fixed at the target site to undergo a deprotection reaction, exposing the active reactive group; and to apply a photocleavage signal to the target site at a fluorescent group cleavage light pulse dose, causing the enzymatically cleavable linker covalently linked to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; The fluorescence sensing control module is configured to switch the fluorescence sensing unit below the target site to a quenched state during the application of the controllable light signal to block the interference of excitation light scattering on fluorescence detection; after confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to a Geiger state to collect the fluorescence signal of the extended base. The base delivery module is configured to deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and to connect the fluorescently labeled base monomers to the primer ends under the catalysis of immobilized enzymes to complete base extension; The cleaning control module is configured to inject cleaning solution into the target site to remove unreacted base monomers, byproducts, and residual enzyme reagents. The AI ​​inference module is configured to identify the type of base being synthesized based on the characteristic fluorescence wavelengths corresponding to different base types; based on the characteristic parameters of the fluorescence signal, it uses a pre-trained inference model to evaluate the synthesis quality of the current base cycle; if the synthesis quality is normal, it proceeds to the next base cycle; if the synthesis quality is abnormal, it adjusts the optical signal parameters and re-executes the local deprotection reaction on the target site; if the synthesis quality fails and cannot be recovered, it marks the target site as an abandoned site and skips subsequent cycles. The timing scheduling module, electrically connected to other modules, is configured to uniformly schedule the input timing of the excitation light source driving module, fluorescence sensing control module, base delivery module, cleaning control module, and AI inference module, achieving site-level independent control and global parallel collaboration.

[0116] In some embodiments, a dose isolation band management module is also included, configured to limit the deprotection light pulse dose to a first dose region and the phosphor removal light pulse dose to a second dose region, wherein a dose isolation band is provided between the first and second dose regions, wherein the relative value range of the light pulse dose in the second dose region is [1, 10], the relative value range of the light pulse dose in the first dose region is [20, 100], and the range of the dose isolation band is (10, 20).

[0117] In this embodiment, the AI ​​inference module is configured to calculate the photon count rate in the four quadrants within the detection window, normalize it to obtain the relative response intensity, and identify the type of base synthesized at the moment based on the relative response intensity of each quadrant. The arrival time of the fluorescence signal photons was recorded by time-correlated single-photon counting technology, and a fluorescence lifetime histogram was constructed. By combining the characteristics of the two dimensions of fluorescence relative response intensity and fluorescence lifetime, four synthesis states were distinguished: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination.

[0118] In this embodiment, the AI ​​inference module is configured to: if the relative response intensity of a certain quadrant exceeds a preset threshold, and the response intensity of the other three quadrants is lower than the cross-interference threshold, determine that the current synthesized base type corresponds to the base in that quadrant. If the response intensity in two or more quadrants exceeds the threshold, or if the fluorescence lifetime histogram shows a bimodal or multimodal distribution, it is determined to be cross-contamination. If the response intensity in all quadrants is below the minimum detection threshold or there is no fluorescence signal, it is determined that it has not been synthesized; If the deviation between the fluorescence lifetime and the standard lifetime of the dye at the primer end is less than the set threshold, it is considered a correct synthesis; If the fluorescence lifetime deviates from the standard lifetime of the dye at the primer end by more than the set range, it is judged as an incorrect synthesis.

[0119] The AI ​​inference model is deployed on the edge computing unit of the optoelectronic co-enzymatic gene synthesis chip.

[0120] Comparative Example 1 This comparative example illustrates the necessity of a dose isolation band mechanism to prevent mode crosstalk. Without a dose isolation band constraint, the light pulse dose from the micro-LED array can be continuously output, including dose values ​​falling within the interval (10, 20).

[0121] When the driving current of the micro-LED has a temperature drift of ±10%, taking the standard parameters (light intensity level 10, pulse width 20ms, nominal relative dose 12) in fluorescence cut-off mode as an example, when the driving current drifts forward by 10%, the actual output dose can reach 13.2, which is still within the continuous dose range and does not trigger the deprotection reaction, but it is close to the upper limit of the low dose region. If we further consider the clock jitter causing the pulse width to be too long, the actual output dose may exceed 14. Without the constraint of the isolation band, the risk of partial photolysis of the protective group cannot be ruled out, thus triggering the false triggering of the deprotection mode.

[0122] In contrast, the method provided in Example 1 uses a dose isolation band mechanism to set the upper limit of the low-dose region to 10 and the lower limit of the high-dose region to 20, maintaining a safety margin of 2 times between them. Under the same ±10% temperature drift conditions, the actual output dose of the fluorescence excision standard parameter (relative dose 5) is at most 5.5, far below the lower boundary of the isolation band of 10, and no mode overflow occurs. The above mechanism achieves reliable isolation between the two photochemical operating modes at the signal level without the need to introduce additional light sources or chemical reagents.

[0123] It is understood that in the above embodiments, the enzymatically cleavable linker is not limited to o-nitrobenzyl derivatives, but may also be coumarin-4-ylmethyl ester or diarylethylene derivatives. Its photolysis wavelength can be finely adjusted to 370nm-390nm to form a slight offset with the 365nm center wavelength of the deprotected light signal, thereby further reducing the risk of crosstalk.

[0124] The fluorescence sensing unit is not limited to a single-photon avalanche diode, but may also be other single-photon sensitive detectors with switchable quenching and Geiger operating modes and switching accuracy that meets the requirements.

[0125] The AI ​​inference model is not limited to small neural networks and threshold classifiers, but may also use decision trees or other lightweight classification models that meet the computing power constraints of chip edge computing.

[0126] The immobilized enzyme is not limited to terminal deoxynucleotidyl transferase, but may also be its mutant, fusion protein or other DNA polymerase with similar catalytic activity.

[0127] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without creative effort within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart control method for a photoelectric co-catalyzed enzymatic gene synthesis chip, applied to at least one base cycle, characterized in that, Includes the following steps: S1. A controllable light signal is applied to the target site of the photoelectric co-enzyme-catalyzed gene synthesis chip with a deprotection light pulse dose, so that the protecting group at the end of the primer fixed at the target site undergoes a deprotection reaction, exposing the active reactive group. S2. During the application of the controllable light signal, the fluorescence sensing unit below the target site is switched to a quenched state to block the interference of excitation light scattering on fluorescence detection. S3. Deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and connect the fluorescently labeled base monomers to the primer end under the catalysis of immobilized enzyme to complete the base extension; S4. Inject cleaning solution into the target site to remove unreacted base monomers, byproducts and residual enzyme reagents; S5. After confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to the Geiger state to collect the fluorescence signal of the extended base and identify the type of base synthesized at present based on the characteristic fluorescence wavelength corresponding to different base types. S6. Apply a photoexcision signal to the target site with a fluorescent group excision light pulse dose to cause the enzymatically cleavable linker arm covalently attached to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; S7. Based on the characteristic parameters of the fluorescence signal, use a pre-trained inference model to evaluate the synthesis quality of the current base cycle; S8. If the synthesis quality is normal, proceed to the next base cycle. If the synthesis quality is abnormal, adjust the optical signal parameters and re-execute the local deprotection reaction on the target site. If the synthesis quality fails and cannot be recovered, mark the target site as an abandoned site and skip the subsequent cycle.

2. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 1, characterized in that, The deprotection pulse dose belongs to the first dose region, the phosphor excision pulse dose belongs to the second dose region, and a dose isolation band is provided between the first dose region and the second dose region; the relative value range of the pulse dose in the second dose region is [1, 10], the relative value range of the pulse dose in the first dose region is [20, 100], and the range of the dose isolation band is (10, 20). The controllable optical signal is ultraviolet light with a wavelength range of 350nm-380nm, an adjustable pulse width of 1μs-100ms, and multi-level quantization of light intensity.

3. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 1, characterized in that, The process of applying a controllable light signal to the target site of the reaction region of the photoelectro-coordinated enzymatic gene synthesis chip with a deprotected light pulse dose includes employing at least one addressing mode among single-point independent addressing, row batch addressing, column batch addressing, or global addressing, wherein the addressing mode is determined according to the site distribution characteristics of the current synthesis task.

4. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 1, characterized in that, The fluorescence sensing unit adopts a 2×2 four-quadrant array structure, with each of the four quadrants configured with a characteristic fluorescence wavelength detection channel corresponding to the four base types. The relative response intensity is obtained by calculating the photon count rate in the detection window of each of the four quadrants and normalizing it. The type of base currently synthesized is identified based on the relative response intensity of each quadrant.

5. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 4, characterized in that, It also includes recording the arrival time of the fluorescence signal photons using time-correlated single-photon counting technology, constructing a fluorescence lifetime histogram, and combining the characteristics of two dimensions, namely fluorescence relative response intensity and fluorescence lifetime, to distinguish four synthesis states: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination. If the response intensity in two or more quadrants exceeds the threshold, or if the fluorescence lifetime histogram shows a bimodal or multimodal distribution, it is determined to be cross-contamination. If the response intensity in all quadrants is below the minimum detection threshold or there is no fluorescence signal, it is determined that it has not been synthesized; If the deviation between the fluorescence lifetime and the standard lifetime of the dye at the primer end is less than the set threshold, it is considered a correct synthesis; If the fluorescence lifetime deviates from the standard lifetime of the dye at the primer end by more than the set range, it is judged as an incorrect synthesis.

6. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 1, characterized in that, The process of adjusting the optical signal parameters and re-performing the local deprotection reaction on the target site includes: adjusting the optical signal parameters by increasing the optical pulse width, increasing the optical intensity level, and changing to a dual-pulse sequence or a continuous pulse sequence, and setting an upper limit on the number of times the local deprotection reaction can be re-performed. If the synthesis quality is still abnormal after re-performing the reaction beyond the upper limit, the synthesis is determined to have failed, the target site is marked as an abandoned site, and a washing solution is injected into the target site in the subsequent base cycle to remove unreacted base monomers, byproducts, and residual enzyme reagents.

7. The intelligent control method for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 1, characterized in that, It also includes the following steps: A global synthesis quality report is generated after each base cycle. After all base cycles are completed, products are collected from reaction sites that are determined to be normal synthesis sites, and waste liquid is discharged from reaction sites that are determined to be waste sites. Normal products and waste products are completely separated in the physical channel.

8. An intelligent control system for a photoelectric synergistic enzymatic gene synthesis chip, characterized in that, include: The excitation light source driving module is configured to apply a controllable light signal to the target site of the reaction region of the photoelectro-co-enzymatic gene synthesis chip at a deprotection light pulse dose, causing the protecting group at the end of the primer fixed at the target site to undergo a deprotection reaction, exposing the active reactive group; and to apply a photocleavage signal to the target site at a fluorescent group cleavage light pulse dose, causing the enzymatically cleavable linker covalently linked to the end of the primer to undergo photolytic cleavage, releasing the fluorescent dye and restoring the free 3'-hydroxyl group; The fluorescence sensing control module is configured to switch the fluorescence sensing unit below the target site to a quenched state during the application of the controllable light signal to block the interference of excitation light scattering on fluorescence detection; after confirming that the controllable light signal has been turned off and the biological reaction system has reached a steady state, the fluorescence sensing unit is switched to a Geiger state to collect the fluorescence signal of the extended base. The base delivery module is configured to deliver base monomer reagents with different fluorescent labels to the target site after the deprotection reaction, and to connect the fluorescently labeled base monomers to the primer ends under the catalysis of immobilized enzymes to complete base extension; The cleaning control module is configured to inject cleaning solution into the target site to remove unreacted base monomers, byproducts, and residual enzyme reagents. The AI ​​inference module is configured to identify the type of base being synthesized based on the characteristic fluorescence wavelengths corresponding to different base types; based on the characteristic parameters of the fluorescence signal, it uses a pre-trained inference model to evaluate the synthesis quality of the current base cycle; if the synthesis quality is normal, it proceeds to the next base cycle; if the synthesis quality is abnormal, it adjusts the optical signal parameters and re-executes the local deprotection reaction on the target site; if the synthesis quality fails and cannot be recovered, it marks the target site as an abandoned site and skips subsequent cycles. The timing scheduling module, electrically connected to other modules, is configured to uniformly schedule the input timing of the excitation light source driving module, fluorescence sensing control module, base delivery module, cleaning control module, and AI inference module, achieving site-level independent control and global parallel collaboration.

9. The intelligent control system for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 8, characterized in that, The AI ​​inference module is configured to calculate the photon count rate in the four quadrants within the detection window, normalize it to obtain the relative response intensity, and identify the type of base synthesized at the moment based on the relative response intensity of each quadrant. The arrival time of the fluorescence signal photons was recorded by time-correlated single-photon counting technology, and a fluorescence lifetime histogram was constructed. By combining the characteristics of the two dimensions of fluorescence relative response intensity and fluorescence lifetime, four synthesis states were distinguished: correct synthesis, no synthesis, incorrect synthesis, and cross-contamination.

10. The intelligent control system for a photoelectric synergistic enzymatic gene synthesis chip as described in claim 9, characterized in that, The AI ​​inference model is deployed on the edge computing unit of the optoelectronic co-enzymatic gene synthesis chip.

Images