DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Priority
Instant application eligible to benefit from the Domestic Priority Data as claimed by applicant on 05/12/2020. This application is a 371 of PCT/JP2020/018902 and EFD was considered as 05/12/2020.
Information Disclosure Statement
IDS has been submitted on 11/20/2022, 02/13/2024 and 09/02/2025 and considered by the examiner. A signed copy of the corresponding 1449 form has been included with this Office action.
Status of Claims
Claims 1-15 are currently pending and are examined on the merits.
Claim 1-15 are rejected.
Claim Objections
Claim 7 is a dependent claim that refers back to a method (claim 6) but identifies itself as a nucleic acid analyzer (an apparatus) considered an improper category. A dependent claim must meet within the same statutory category as the independent claim it refers to.
Applicants are required to amend the claim by stating: “The nucleic acid analysis method according to claim 6…”
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
Under MPEP 2181, section I, a limitation invokes 112(f) if it meets the following:
Uses "means" or a generic placeholder (nonce word) for structure.
The term is modified by functional language.
The term is not modified by sufficient structure to perform the function.
Based on the three-prong analysis for evaluating 35 U.S.C. 112(f) (means-plus-function) limitations, as outlined in MPEP § 2181, the claim 1 triggers 112(f) issues, specifically for the limitations “base prediction unit,” “registration unit,” and “extraction unit.”
Base prediction unit
Prong A (Generic placeholder): The term “unit” is a nonce word used as a substitute for “means”.
Prong B (Functional language): The unit is modified by functional language: “configured to perform base prediction using, as an input”
Prong C (Insufficient structure): The term “unit” does not support specific, known physical structure, such as “sensor”. The description “base prediction unit” describes what it does (predicts) rather than what it is.
Therefore, 112(f) Invoked.
Registration unit
Prong A (Generic placeholder): "Unit" is a nonce word.
Prong B (Functional language): Modified by “configured to perform registration of the plurality of images relative to a reference image”.
Prong C (Insufficient structure): No specific structure (e.g., specific hardware) is disclosed within the claim term itself to perform the registration.
Therefore, 112(f) Invoked.
Extraction unit
Prong A (Generic placeholder): “Unit” is a nonce word.
Prong B (Functional language): Modified by “configured to extract a spot from the plurality of images”.
Prong C (Insufficient structure): “Extraction unit” is a pure functional term for a software.
Therefore, 112(f) Invoked.
However, in the specification applicant discloses for each of the units (Base prediction, registration and extraction) with sufficient structure”. The specification paragraphs [0056], [0087], [0091], [0017] describes the what each unit defines with sufficient structure. Because the claim, when read in view of the specification, connects the “units” to a specific structure, it satisfies the requirements for definiteness and structure. So, claim 1 is not objected under U.S.C. 112(f) (FP 7.30.03, 7.30.05) (MPEP § 706.03(e) and § 2181).
Claim 3 introduces a new term “predictor,” but it fails to overcome the 112(f) issue.
Prong 1: Is there a substitute for “means”? (Yes)
The claim replaces “unit with “predictor”. “Predictor” is generally treated as nonce term. Much like “controller” or “processor.” It identifies a functional role (something that predicts) rather than a specific physical structure.
Prong 2.: Is the term modified by functional language.? (Yes)
The predictor is defined by its ability to “perform supervised learning.”
Prong 3.: Does the claim lack sufficient structure to perform the function? (Yes)
“Supervised learning” is a broad category of machine learning (e.g., linear regression, SVMs, neural networks, random forest). The claim does not specify how the predictor is structured (e.g., the architecture of the neural network) to transform the “peripheral pixels” into a “base prediction.” Simply stating it can perform “supervised learning” is a functional requirement, not structural description.
Therefore, the claim still invokes 112(f).
However, in the specification applicant discloses about predictor (paragraph [0135]; Figure 18) and supervised learning (paragraph [0118], [0163], figure 16) with sufficient structure”. Because the claim, when read in view of the specification, connects the “units” to a specific structure, it satisfies the requirements for definiteness and structure. So, claim 1 is not objected under U.S.C. 112(f) (FP 7.30.03, 7.30.05) (MPEP § 706.03(e) and § 2181).
Claim 5 introduces a “plurality of base prediction units” but still invokes 112(f) issue.
Prong 1.: Is there a substitute for “means”? (yes)
The claim uses the term “units.” As analyzed previously, “unit is a nonce term- a generic placeholder that lacks physical or structural meaning.
Prong 2.: Is the term modified by functional language? (Yes)
The Plurality of units is modified by the requirement that they collectively “predict a base based on prediction results of the plurality.” This describes a functional hierarchy rather than the physical composition of the units.
Prong 3.: Does the claim lack sufficient structure to perform the function? (Yes)
The claim describes multiple units feeding into final results but provides no structural detail for the units themselves. Simply multiplying the number of units is not sufficient to move the claim out of 112(f) invocation. Therefore, the claim invokes 112(f).
However, in the specification applicant discloses about base prediction unit (paragraphs [0056], [0017]) with sufficient structure”. Because the claim, when read in view of the specification, connects the “units” to a specific structure, it satisfies the requirements for definiteness and structure. So, claim 1 is not objected under U.S.C. 112(f) (FP 7.30.03, 7.30.05) (MPEP § 706.03(e) and § 2181)
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title
Claims 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 2A, Prong 1
In accordance with MPEP § 2106, the instant claims 1-5, are drawn to a process (method), claims 6-9 and 10-15 are drawn to a method, and therefore are found to recite statutory subject matter (Step 1: YES). The instant claims are then analyzed to determine if the claims recite any concepts that equate to an abstract idea, law of nature or natural phenomenon (Step 2A, Prong 1). The instant claims recite the following limitations that equate to an abstract idea:
Claim 1 recites
Predicting a base” based on “feature data” extracted from image data (Mental process or mathematical concept). These steps represent a series of observations, evaluations, and judgements that can be performed in the human mind or through mathematical calculations.
Perform registration of the plurality of images relative to a reference image; and an extraction unit configured to extract a spot from the plurality of images. (Mathematical concept) The registration and extraction processes rely on algorithms, calculations, and mathematical transformations.
Claim 3 recites “base prediction unit is implemented by a predictor capable of performing supervised learning”. (Mathematical concept) Machine learning algorithms (including supervised learning predictors) are classified as mathematical concepts (abstract ideas) because they typically use algorithms to process data.
Claim 4 recites “the base prediction unit receives, …… an image in at least one cycle selected from a previous cycle and a next cycle as an input”. (Mental process) The act of comparing data from different point in time (“a previous cycle” or “next cycle” to inform a prediction is a logic-based method of data analysis. Human routinely perform this type of contextual analysis.
Claim 5 recites “a plurality of the base prediction…. a base is predicted ……. the plurality of base prediction units.” (Mental process) The concept of obtaining multiple predictions and combining them to reach a final conclusion is a fundamental human activity and mathematical method that can be performed in the mind or using pen and paper.
Claim 6 recites
Executing a colony position determining stage and a base sequence determining stage. (Mathematical concept) These stages involve determining a value (e.g., coordinates, DNA/RNA sequence codes), which is a mathematical concept.
Registration processing of registering the plurality of images. (Mental process) This can be performed in the human mind (e.g., observing, evaluating, classifying images).
Claim 9 recites “the biologically related substance is determined by extracting a spot from the plurality of images captured at temporally different timings. (Mental process) The determination of a position by “extracting a spot” from images captured at “temporarily different timings” is logical method of data filtering and spatial identification.
Claim 10 recites “the machine learning method comprising: a first base prediction step of generating a ……training data generation step of generating first training……a predictor updating step of updating …. using the second training data” (Mental process). The steps of generating a prediction, aligning with a reference and updating a parameter are fundamental mathematical and logical operations.
Claim 13 recites “the second training data is determined based on information of at least one of a signal intensity and likelihood, and an ……the reliability. (Mental process) Including data from previous or next cycle to inform a prediction is a method of mathematical modeling of time sequence data. Identifying patterns across a sequence of events to improve a prediction is a logical process that can be performed in the human mind.
Claim 14 recites “the training data updating step, ……is added to the first training data.” (Mental process). The step of identifying data present in one set but not another and adding it to the first is a basic logical operation. This represents a mental process for organizing information’s that a human can perform.
Claim 15 recites “a predictor re-updating step of updating a parameter of the base predictor using the first training data updated in the training data updating step.” (Mental process) The predictor reupdating step is mathematical optimization of a model parameters. The concept of learning from a refined dataset to improve a previous conclusion is a fundamental mental process.
Claim 11 and 12 provides information with further limitation without active steps.
As such claims 1-15 recite an abstract idea (Step 2A, Prong 1: YES).
Step 2A, Prong 2
Claims found to recite a judicial exception under Step 2A, Prong 1 are then further analyzed to determine if the claims as a whole integrate the recited judicial exception into a practical application or not (Step 2A, Prong 2). Specifically, the claims recite the following additional elements:
Claim 1 recites “the base prediction unit receives, as an input, an image including peripheral pixels around a position of the spot extracted from the plurality of images.
Claim 2 recites “plurality of images are obtained by detecting, by a sensor, a plurality of types of luminescence from a plurality of types of fluorescent substances”.
Claim 4 recites “the base prediction unit receives, in addition to an image in a cycle to be predicted, an image in at least one cycle selected from a previous cycle and a next cycle as an input.”
Claim 6 recites “the base sequence determining stage, the base predictor receives, as an input, …. peripheral pixels …. extracted…. feature data.”
Claim 7 recites “the plurality of images are obtained by detecting, by a sensor, …. related substance.”
Claim 8 recites “the base sequence determining stage, the base predictor receives, …. peripheral pixels …… plurality of images.”
Claim 10 recites “performing base prediction using, as an input, a plurality of images obtained by detecting luminescence from a biologically related substance.”
The additional elements for claim 2 and 7 “the plurality of images are obtained by detecting, by a sensor”, although it adds sensors regarding how the luminescence is detected, the core fundamentals of the claim still remain the algorithmic determination of a colony position and the prediction of a base on the feature data.
The additional elements for claim 1, 4, 6, 8 and 10 “receiving input or peripheral pixels” in the process of predicting a base. The steps of receiving “input” or “peripheral pixels” are categorized as merely collecting data and feeding it into a mathematical or mental process. Under Alice Corp. v. CLS Bank Int'l, simply gathering and crunching data on a general-purpose computer or analyzer does not transform an unpatentable abstract idea into a patent-eligible application. The act of receiving pixel inputs and predicting a base is well-understood, routine, and conventional in sequence by synthesis technology. Because these additional elements do not reflect an improvement in computer/or analyzer functioning or physical technology, the claim fails to integrate the exception into a practical application and serves as being merely an insignificant, routine, or conventional post-solution activity and used an input for the judicial exception. Therefore, these limitations are mere data gathering or analyzing activities using a conventional computing/sequence analyzer system. As set forth in MPEP 2106.05(g), mere data gathering and analyzing activity has been identified by the courts as insignificant extra-solution activity that does not provide a practical application.
There are no limitations that indicate that the nucleic acid analyzer consisting of base prediction unit, image registration unit require anything other than a conventional sequencing by synthesis (SBS) technology. As such, these limitations equate to mere instructions to implement the abstract idea on a generic computer/analyzer that the courts have stated does not render an abstract idea eligible in Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983. See also 573 U.S. at 224, 110 USPQ2d at 1984.
The above recited additional elements do not provide a practical application of the recited judicial exception. As such, claims 1-15 are directed to an abstract idea (Step 2A, Prong 2: NO).
Step 2B
Claims found to be directed to a judicial exception are then further evaluated to determine if the claims recite an inventive concept that provides significantly more than the judicial exception itself (Step 2B). The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the claims recite additional elements that equate to mere instructions to apply the recited exception in a generic computing and/or sequencing by synthesis (SBS) environment or well-understood, routine and conventional activity.
As discussed above, there are no additional limitations to indicate that the claimed analyzer requires anything other than generic computer components/or sequence analyzer in order to carry out the recited abstract idea in the claims. Claims that amount to nothing more than an instruction to apply the abstract idea using a generic computer do not render an abstract idea eligible. Alice Corp., 573 U.S. at 223, 110 USPQ2d at 1983. See also 573 U.S. at 224, 110 USPQ2d at 1984.
Furthermore, the additional elements recited in the claims 2 and 7 “the plurality of images are obtained by detecting, by a sensor”, the sensor is a typical sensor in the ordinary field of use of providing data and amount to well-understood, routine and conventional process for the sequencing by synthesis. (Garrido-Cardenas et al. -Abstract) Garrio-Cardenas et al. describes in his review the conventional use of sensors in sequencing technology. Also, additional elements such as obtaining, receiving and outputting information in claim 1, 2, 4 and 7 is insignificant, extra-solution activity that is not significantly more than the judicial exceptions. As such, the combination of additional elements recited in the claims is well-understood, routine and conventional.
The additional elements do not comprise an inventive concept when considered individually or as an ordered combination that transforms the claimed judicial exception into a patent-eligible application of the judicial exception. Therefore, the claims do not amount to significantly more than the judicial exception itself (Step 2B: No). As such, claims 1-15 are not patent eligible under 35 U.S.C. §101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1, 3, 4, 10 and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. (US 20180274023 A1) in view of Kermani et al. (US 20150169824 A1).
Regarding claim 1:
Belitz et al. discloses:
A nucleic acid sequencing /image analysis system for performing sequencing by synthesis analysis using image data obtained from fluorescently labeled nucleic acid clusters disposed on a substrate (paragraph 0238-0240, 0028-0030, 0062, 0067; Claim 73 and 88; abstract; Fig 3) suggesting limitations of “plurality of images obtained by detecting luminescence … on a substrate”
Registering images relative to a reference image (Paragraph 0036, 0037, 0090, 0091) which suggest limitation of “perform registration of the plurality of images relative to a reference image”.
Extracting feature/intensity data from image regions (Paragraph 0151,0038; claim 73 and 88) reads the limitation of “an extraction unit …. plurality of images”.
Determining nucleotide bases from the extracted signal/image information. (Paragraph 0013, 0014, 0036, 0037) suggesting the limitation of “extracts …. a base based on the feature data”.
Belitz et al. therefore teach:
A nucleic acid analyzer comprising: A registration unit configured to perform registration of the plurality of images relative to a reference image; and an extraction unit configured to extract a spot from the plurality of images.
However, Belitz et al. do not explicitly disclose that the base prediction unit:
Receives an image including peripheral pixels around the extracted spot.
Extract feature data using machine learning/deep learning (prediction method) feature extraction, and predicts a base based on learned feature (image).
Kermani et al. discloses
A DNA sequencing base caller using machine learning techniques. (Paragraphs 0032, 0058)
Using image data including neighboring/peripheral pixels surrounding sequencing spot/clusters. (Abstract, Paragraphs 0036,0048,0077,0157--159, 0209, 0069, 0210) reads the limitation of “the base prediction unit receives, …. spot extracted from the plurality of images”.
Extracting feature from the image data using machine learning models. (Paragraphs 0033,0034,0058; Figure 1) reads the limitation of “a base prediction unit configured to perform base prediction”.
Predicting nucleotide bases from the extracted features. (Paragraph 0032) suggests limitation of “predicts a base based on the feature data.”
Therefore, it would have been obvious to a person skill in the art (PHOSITA) at the time of the invention to modify an analyzer (Belitz et al.) (a standard SBS (sequencing by synthesis) machine by Illumina) with the specific peripheral pixels use and feature based prediction method taught by Kermani et al. in order to improve sequencing accuracy, signal determination, noise robustness and automated feature extraction from sequencing image data. (MPEP 2143 and KSR Int’l Co. v. Teleflex., 550 U.S. 398 (2007).
A PHOSITA would be motivated by the market/known need for improvement accuracy in high density sequencing by synthesis (SBS) to modify the existing sequencing technologies. Using a window of peripheral pixels (Kermani et al.) in a feature extraction algorithm (Kermani et al.) would allow to better differentiate true signals from background noise and yield a predictable success of the invention.
The combination of these references teaches every element of the claim: a registration and extraction system (Belitz et al.) that feeds localized spot images including peripheral pixels (Kermani et al.) into base prediction unit that extracts feature data to make a base call. Consequently, the claim is rejected under 35 U.S.C. 103.
It would be an exception of success of using all these cited arts together because they are all in the same technology (DNA sequencing), addressing the problem with base prediction accuracy.
Regarding claim 3:
Claim 3 is a dependent claim off claim 1 and Belitz et al. discloses most of the limitations in claim 1. However, Belitz et al. do not explicitly disclose: “the base prediction unit is …. capable of performing supervised learning”
Kermani et al. discloses sequencing image analysis system employing machine learning predictors trained using labeled sequencing data. (Paragraph 0032). The reference teaches supervised learning approaches in which training data having known nucleotide identities are used to train prediction/classification models for sequencing base calling. He further discloses deep-learning -based sequencing image analysis using supervised training techniques.
The primary motivation is the enhancement of base-calling accuracy in the presence of noise and signal decay. Supervised learning model are specifically designed to handle complex, non-linear relationship in image data that traditional threshold -based algorithms may miss. Implementing a supervised learning model is a predictable application of known machine learning techniques to the field of genomics. Using labeled datasets to train a predictor was a well-known and widely adopted strategy in image analysis at the time of the invention.
A PHOSITA would recognize that supervised learning is the standard methodology for training the image data and statistical classifiers already suggested for use in high throughput sequencing system. Therefore, it would have been obvious to one of ordinary skill in the art at the time the invention was made to modify the base calling/image -analysis system of Belitz et al. to implement the base prediction unit using supervised -learning predictors as taught by Kermani et al. in order to improve sequencing accuracy, adaptability, signal classification performance, and automated feature learning from sequencing image data. (MPEP 2141 and 2143 and KSR Int’l Co. v. Teleflex., 550 U.S. 398 (2007).
Regarding claim 4:
Claim 4 is a dependent claim off claim 1 and Belitz et al. discloses most of the limitations in claim 1. Also, Belitz et al. discloses sequencing image analysis using image information from multiple sequencing cycles, including adjacent cycles, to improve nucleotide determination accuracy (phase correction) (Claim 81, 86; Paragraph 0046, 0207, 0212, 0213, 0215). The references teach analyzing sequencing signals across prior and subsequent cycles to compensate for phasing, pre-phasing, noise and signals overlap effects. The reference further discloses using temporally adjacent cycle image information when performing sequencing analysis and base calling.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of the filling date to modify the sequencing image processing and base calling system of Kermani et al. to receive image information from previous and/or next sequencing cycles as input, as taught by Belitz et al. in order to improve sequencing accuracy by accounting for phasing effects, temporal signal variation, and inter-cycle sequencing dependencies.
The motivation to combine is supported by the references shared objective of improving sequencing base calling accuracy using information from multiple sequencing cycles. (MPEP 2141 and 2143 and KSR Int’l Co. v. Teleflex., 550 U.S. 398 (2007).
Regarding claim 10:
Belitz’s and Kermani disclose:
Generating sequencing base prediction results from sequencing image data (Belitz)
Training machine learning predictors/classifiers using sequencing image features (Kermani; Paragraph 0053, 0083, 0201)
Updating prediction model parameters using training data (Kermani; Paragraph 0053, 0083, 0201)
Iteratively improving sequencing prediction performance (Kermani; Paragraph 0053, 0083, 0210)
Kermani further discloses deep-learning sequencing image analysis methods using neural network predictors trained from sequencing image datasets and reference sequence information. Iterative machine-learning training methods in which initial training data are generated trained and updated.
Therefore, the references teach: Generating sequencing predictions from image data, compare predicted sequences to reference sequences, generating supervised training information from alignment/comparison results, updating neural network parameters using the training data, and re-performing sequencing prediction using updated model parameters.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of EFD to modify the sequencing machine-learning methods of Kermani teaching to iteratively update training datasets using subsequently generated training data, in order to improve sequencing prediction accuracy, reduce labeling noise, improve convergence of the learning process, and iteratively refine the base predictor.
Regarding claim 11:
Claim 11 is a dependent claim off claim 10 and all of the limitations can be mapped using prior art from claim 10 rejection.
The limitations in claim 11 are the same as limitations in claim 4. Therefore, the same claim mapping (for claim 4) can be applied to claim 11 for the rejection.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of the EFD to modify the machine learning sequencing prediction methods to provide image information from previous and/or next sequencing cycles as input to the base predictor, in order to improve sequencing prediction accuracy by incorporating temporal sequencing context and compensating for sequencing cycle signal variation.
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. in view of Kermani et al. as applied to claims 1, 3, 4, 10 and 11 above, and further in view of Feng et al.(US 7329860 B2)
Belitz et al. in view of Kermani et al. are applied to claims 1, 3, 4, 10 and 11 above.
Claim 2 is a dependent claim off claim 1 recites “the plurality of images are obtained by detecting, by a sensor, a plurality of types of luminescence from a …. and an optical path to the sensor for detection.”.
However, Belitz et al. and Kermani et al. do not explicitly disclose: “the plurality of images are obtained by detecting, by a sensor …. sensor for detection.”
Feng et al. teaches the following:
Multiple luminescence types (Paragraph 0030), and Multiple fluorescence substances Paragraph 0088, 0101). reads limitation of “plurality of …. of fluorescent substances”
Detection by sensor (Figure 12, 14: Paragraph 0085, 0088, 0101) reads to detecting by a sensor.
Different via optical path (Paragraph 0034,0038,0041, 0042) and different via sensor (Figure 12, 14; Paragraph 0085, 0088, 0101) reads to “the plurality of types of luminescence …. sensor for detection and an optical path …sensor for detection.”
Feng et al. discloses confocal systems in which different fluorescence emissions are separated by wavelength -selective optics and detected using different sensor path/channels for imaging nucleotides in tissues. (Paragraph 0146, 0165, 0146). The reference teaching optical spiriting system, multiple detector arrangement, and distinct, optical paths for different luminescence signals generated from fluorescently labeled nucleic acids on a tissue specimen.
Therefore, it would have been obvious to one of ordinary skill in the art at the time the invention was made to modify the sequencing image analysis system of Belitz et al. to employ the multichannel fluorescence detection arrangement taught by Feng et al. in order to improve discrimination of fluorescent signals corresponding different nucleotide incorporations and improve sequencing accuracy.
The primary motivation for this modification is to increase throughput and sensitivity. Utilizing distinct optical path or sensors (massively parallel detection) allows for parallel detection of multiple signals, which a PHOSITA would recognize the key factor for high-speed sequencing with accuracy.
Employing separate sensors or paths is a known engineering solution to resolve cross-talk between fluorescence substances with overlapping spectra. A POSITA would recognize that such a configuration, as taught by Feng et al., and for the optimization needs in image processing, would lead to higher quality image data for the base prediction unit.
The combination of these elements yields a high expectation of success. Using separate sensors/paths to resolve overlapping fluorescence spectra is a well-established solution, meaning a PHOSITA would confidently expect it to work without unpredictable results.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. in view of Kermani et al. as applied to claims 1, 3, 4, 10 and 11 above and further in view of Holm et al. (US 2019/0355364 A1)
Belitz et al. in view of Kermani et al. are applied to claims 1, 3, 4, 10 and 11.
Claim 5 is a dependent claim off claim 1 and recites “base is predicted based on prediction results of the plurality of base prediction units”
However, Belitz et al. do not explicitly disclose: “a base is predicted based on prediction results of the plurality of base prediction units.”
Holm et al. discloses ensemble machine learning systems employing multiple prediction models/classifiers and combining outputs from the plurality of prediction models to generate a final prediction result. (Paragraph 0055-0062) which suggest the limitation of final prediction from combined predicted results from plurality of prediction units. The reference teaches combining prediction outputs from multiple classifiers using ensemble decision logic to improve prediction accuracy and robustness.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of effective filing date to modify the sequencing base calling system of Belitz’s and Kermani et al. to employ multiple base prediction units whose prediction outputs are combined to generate a final nucleotide prediction, as taught by Holm et al., in order to improve sequencing prediction accuracy, reduce classification noise, improve robustness to image artifacts, and increase confidence in nucleotide determinations.
The motivation to combine is supported by the references shared objective of improving prediction accuracy using multiple classifier and aggregated prediction techniques. Ensemble prediction methods were well known prior to EFD for improving classification reliability in machine learning systems. (MPEP 2141 and 2143 and KSR Int’l Co. v. Teleflex., 550 U.S. 398 (2007).
Claim 6, 8 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. in view of Kermani et al. as applied to claims 1, 3, 4, 10 and 11 above and further in view of Kircher et al. (BMC Genomics 2011, 12:382)
Belitz et al. in view of Kermani et al. are applied to claims 1, 3, 4, 10 and 11.
Regarding claim 6:
Belitz et al. discloses
A nucleic acid sequencing /image analysis system for performing sequencing by synthesis analysis using image data obtained from fluorescently labeled nucleic acid clusters disposed on a substrate (paragraph 0238-0240, 0028-0030, 0062, 0067; Claim 73 and 88; abstract; Fig 3) suggesting limitations of “plurality of images obtained by detecting luminescence … on a substrate”
Registering images relative to a reference image (Paragraph 0036, 0037, 0090, 0091) which suggest limitation of “perform registration of the plurality of images relative to a reference image”.
Extracting feature/intensity data from image regions (Paragraph 0151,0038; claim 73 and 88) reads the limitation of “an extraction unit …. plurality of images”.
Determining nucleotide bases from the extracted signal/image information. (Paragraph 0013, 0014, 0036, 0037) suggesting the limitation of “extracts …. a base based on the feature data”.
Using surrounding/background pixels around cluster locations. (Paragraph 0004 0006, 0044, 0054, 0066, 0032)
Acquiring a plurality of sequencing-cycle fluorescence images. [Paragraph 0003, 0092]
Determining cluster positions from the image data. [paragraph 0156]
Therefore, Belitz’s disclose a sequencing analysis workflow including: Image registration processing, cluster/spot extraction processing, cluster position determination and base determination from sequencing image data.
Kermani et al. discloses machine-learning and deep learning sequencing image analysis methods where he teaches
Extracting image regions surrounding sequencing colonies/clusters (Paragraph 0036, 0048) including neighboring/peripheral pixels in the image input (Paragraph 0048) suggesting the limitation of “executing a colony position determining stage ….. in the colony position determining stage”
Extracting learned feature representations from the image data using neural networks (Paragraph 0069, 0157,0158, 0159, 0210, 0209,) and predicting nucleotide bases from the extracted image features (Paragraph 0032) suggesting the limitation of “the base predictor receives, as an input, an image including ….. extracts feature data of the image, and predicts a base based on the feature data”.
However, Belitz and Kermani do not explicitly disclose: “The base predictor receives, as an input, an image including peripheral pixels around the colony position extracted from the plurality of images, extract feature data of the image, and predicts a base based on feature data”
Kircher et al. discloses sequencing workflows in which sequencing image registration (Figure 1) and cluster position (Figure 1; pg. 10, c2, middle) determination are performed before downstream sequencing analysis and base calling operations (Figure 1; pg. 3, c1 middle) The reference teaches multi-stage sequencing image processing pipelines including image registration, feature localization, and subsequent sequence determination.
Therefore, it would have been obvious to one ordinary skill in the art at the time of effective filing date to modify the sequencing image analysis method of Belitz’s to employ the machine learning feature extraction and prediction method taught by Kermani et al., while organizing the sequencing workflow into registration/colony position determination stages followed by sequence determination stages as taught by Martin Kircher et al., in order to improve sequencing accuracy, robustness to noise, and automated feature extraction from sequencing image data.
The motivation to combine is supported by references common goal of improving sequencing base calling accuracy through improved image registration, colony localization, and machine-learning based image analysis. It would be an exception of success of using all these cited arts together because they are all in the same technology, addressing the similar problem.
Regarding claim 8, all of the limitations are taught by prior arts that are used for the rejection of claim 6 including the limitation of “a set including a plurality of images captured at temporally different timings.”
Belitz’s discloses sequencing image processing using image information acquired at different sequencing cycles/times [Paragraph 0029, 0057, 0099]. The reference teaches analyzing image data from temporally different sequencing cycles, including neighboring cycles, to compensate for phasing, pre phasing, signal carryover, and sequencing noise effect.
Therefore. It would have been obvious to one of ordinary skill in the art at the time of EFD to modify the sequencing image analysis method of Belitz’s to provide to the base predictor, a set including multiple temporally distinct sequencing images surrounding a colony position, in order to improve sequencing prediction accuracy by utilizing temporal sequencing context and compensating for sequencing -cycle signal variation.
Regarding claim 9, all of the limitations are taught by prior arts that are used for the rejection of claim 6 including the limitation of “a position of the …… is determined by extracting a spot from the plurality of images captured at temporally different timings.”
Belitz’s discloses sequencing image analysis using image data acquired from multiple temporally distinct sequencing cycles. [Paragraph 0008, 0011, 0030] suggesting the limitation of “a position of the …… is determined by extracting a spot from the plurality of images captured at temporally different timings.” The reference teaches determining sequencing feature/cluster locations using image information from multiple sequencing cycle to improve localization accuracy and compensate for phasing, signal variation, and imaging noise. He further discloses extracting sequencing image features/spots using image patches obtained from multiple sequencing images and combining information from multiple image frames/cycles to improve feature localization and sequencing analysis accuracy.
Therefore, it would have been obvious to one of ordinary skill in the art at the time of effective filing date to modify the sequencing image method of Belitz’s to determine colony positions using spot extraction from multiple images captured at temporally different timings, in order to improve colony position accuracy, reduce noise sensitivity and compensate for sequencing -cycle signal variability.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. in view of Kermani et al., Martin Kircher et al. as applied to claims 1, 3, 4, 10, 11, 6, 8, and 9 above and further in view of Feng et al. (US 7329860 B2)
Belitz et al. in view of Kermani et al., and Martin Kircher et al. are applied to claims 1, 3, 4, 10, 11, 6, 8, and 9.
Claim 7 is a dependent claim off claim 6 and most of the limitation is mapped using prior art from claim 6 rejection except the “plurality of images … detecting, by a sensor, a …. and an optical path to the sensor for detection”
Belitz et al. in view of Kermani et al., and Martin Kircher et al. does not explicitly teach “the plurality of images are obtained by detecting, by a sensor”
Feng et al. teaches the following:
Multiple luminescence types (Paragraph 0030), and Multiple fluorescence substances Paragraph 0088, 0101) reads the limitation of “plurality of …. of fluorescent substances”
Detection by sensor (Figure 12, 14: Paragraph 0085, 0088, 0101) reads the limitation of “detecting by a sensor.”
Different via optical path (Paragraph 0034,0038,0041, 0042) and different via sensor (Figure 12, 14; Paragraph 0085, 0088, 0101) reads to “the plurality of types of luminescence …. sensor for detection and an optical path …sensor for detection.”
Feng et al. discloses confocal systems in which different fluorescence emissions are separated by wavelength -selective optics and detected using different sensor path/channels for imaging nucleotides in tissues. (Paragraph 0146, 0165, 0146). The reference teaching optical spiriting system, multiple detector arrangement, and distinct, optical paths for different luminescence signals generated from fluorescently labeled nucleic acids on a tissue specimen.
Therefore, it would have been obvious to one of ordinary skill in the art the time the invention was made to modify the sequencing image -analysis method of Belitz to employ the multi channels fluorescence detection architectures taught by Wenyi Feng et al. in order to improve sequencing signal separation, fluorescence discrimination, and sequencing accuracy.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al. in view of Kermani et al. as applied to claims 1, 3, 4, 10, and 11 above and further in view of Mikolajezyk et al. (Data augmentation for improving deep learning in image classification problem: IEEE Xplore: 21 June 2018)
.
Belitz et al. in view of Kermani et al. are applied to claims 1, 3, 4, 10, and 11.
Belitz et al. in view of Kermani et al. does not teach augmentation of several data generation steps such as “at least one of the first training … added to at … second training data.”
Mikolajezyk et al.. disclose machine-learning data augmentation techniques in which additional training images are generated by applying image processing transformations to existing training images. The reference teaches:
Generating augmented image datasets. (Figure 7; Abstract)
Adding transformed images to training datasets to improve machine-learning robustness and prediction accuracy. (Abstract)
Data augmentation for improving deep learning in image specification problem. (Abstract)
The above teaching reads to claim limitation of “at least one of the first training … added to at … second training data.”
Mikolajezyk et al. further teaches augmenting training datasets during iterative training data generation processes by adding processed/transformed versions of existing training images.
The motivation is to improve model generalization and prevent overfitting. In genomic imaging, slight variation in focus, rotation, or signal intensity can occur. By adding processed versions of original images to the training set (augmentation), the PHOSITA would ensures the predictor remains accurate despite these variations at the time of the EFD.
Data augmentation is a standard industry practice in supervised learning. Applying it to the specific first or second training data generation steps of an iterative sequencing method is a predictable application of known machine learning principles to achieve a more robust base caller.
Adding a preprocessed image to the training set is a routine way to artificially increase the size and diversity of a training dataset when labeled biological samples are limited or expensive to obtain. Because the prior art specifically teaches generating additional training samples by applying image processing to original images, the additional limitation of claim 12 does not provide a patentable distinction over the combined teachings of the cited references.
Claim 13, 14 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Belitz et al., and Kermani et al. as applied to claims 1, 3, 4, 10, and 11 above and further in view of Li et al. (Molecules 2018, 23(8), 1923)
Belitz et al., in view of Kermani et al. are applied to claims 1, 3, 4, 10, and 11.
Regarding claim 13:
Belitz et al., in view of Kermani et al. does not teach confidence level generation steps such as “reliability of an image …. and likelihood, and an image to be used for at …. reliability.”
Li et al. discloses machine-learning systems that determine confidence/reliability values for training samples and prediction outputs using likelihood scores, signal strengths, confidence metrics, or classifier probability outputs (pg. 14, middle; pg. 9, middle) which suggests the limitation of “reliability of an image …. and likelihood”. He also discloses: Evaluating reliability of image data based on confidence or likelihood information and selecting training data based on confidence thresholds. (pg. 3 middle)
Additionally, sequencing systems prior to EFD routinely employed signal intensity and quality/confidence metrics (for example, Q-scores, intensity thresholds) to determine reliability of sequencing data and filter low quality sequencing signals.
Therefore it would have been obvious to one of ordinary skill in the art before the EFD to modify the sequencing machine learning methods Belitz’s and Kermani to determine reliability of sequencing training images based on signal intensity and /or likelihood information and select training images based on the determined reliability, as taught by Li et al., in order to improve training data quality, reduce noise contamination, improve machine-learning convergence, and improve sequencing prediction accuracy.
Regarding claim 14:
Belitz et al., in view of Kermani et al. does not teach filtering steps such as “the training data updating step, …. included in the first training data, is added to the first training data.”
A person of ordinary skill in the art (PHOSITA) before the effective filing Date (EFD) would have found it obvious to modify the iterative sequencing machine learning methods of Kermani et al. Specifically, it would have been a routine optimization and conventional filtering procedure to augment the first training data by identifying and adding newly generated second training data not already present therein.
This combination represents the mere application of a known data-filtering technique to a known method ready for improvement. (MPEP 2143) It yields predictable results, predictably functioning to expand training data sets, improve training diversity, enhance machine learning convergence, and increase sequencing prediction robustness and accuracy.
Regarding claim 15:
Belitz et al., in view of Kermani et al. does not teach “a predictor reupdating step of updating a parameter of the base predictor ……. data updating step.”
Li et al. discloses deep-learning sequencing analysis methods using neural-network predictors trained using sequencing image datasets and reference sequence alignment information. The reference teaches: Generation of updated prediction results using updated parameters. (pg. 5, bottom) suggesting the limitation of “a predictor reupdating step of updating a parameter of the base predictor” Li further discloses cyclic retaining workflows in which updated training datasets are fed back into the model-training pipelines to perform additional parameter optimization and model refinement. (pg. 5, bottom; pg. 6, bottom; pg. 8, top)
It would have been obvious to a person of ordinary skill in the art (PHOSITA) at the time of the effective filing date (EFD) to iteratively update and retain a base predictor using refined training data taught by Li et al. Modifying sequencing machine learning methods in this manner directly achieves faster convergence, higher sequencing prediction accuracy, greater robustness, and improved model refinement.
Both references share a stated objective of improving prediction performance through iterative retraining and updated datasets. Furthermore, applying a known machine learning optimization technique such as repeated retraining following data refinement to the systems of the references would have been obvious to a POSITA to predictably improve overall model accuracy prior to the EFD.
Conclusion
No claims are allowed.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ARSHAD KHAN whose telephone number is (571)272-9812. The examiner can normally be reached Mon-Fri-7:30-5:00 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Larry Riggs can be reached at 5712703062. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/A.H.K./Examiner, Art Unit 1686
/LARRY D RIGGS II/Supervisory Patent Examiner, Art Unit 1686
Prosecution Insights