DETAILED ACTION
Examiner Remarks
In light of Applicant’s arguments and amendments made in the Remarks submitted on 08/07/2026, Examiner has withdrawn the 112(f)-claim interpretation and the previous 101 rejection after finding that the amended claim limitations recite an improvement in technology.
Response to Arguments
Applicant’s arguments with respect to the independent claims have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged by Applicant’s argument as found in the Remarks submitted on 08/07/2026.
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 .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 07/06/2026, 07/30/2026, and 08/07/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
Specification
The specification is objected to as failing to provide proper antecedent basis for the claimed subject matter. See 37 CFR 1.75(d)(1) and MPEP § 608.01(o). Correction of the following is required: newly added claims 22-24 recite the claim element of “computer storage media” which is not found in the specification.
Claim Rejections - 35 USC § 103
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.
Claims 1-7, 14, 16 and 20-24 are rejected under 35 U.S.C. 103 as being unpatentable over Kumaran et al., CA 3051463-A1(“Kumaran”) in view of Volk, Michael Jeffrey, et al. "Biosystems design by machine learning." ACS synthetic biology 9.7 (2020)(“Volk”) and in view of Stajdohar et al., US 2024/0233874 Al(“Stajdohar”)
Regarding claim 1, Kumaran teaches a computer-implemented method for training a machine learning model with specialized biologic data in an AI-guided analytic platform for development of a biologic synthesis process, the method comprising:
collecting multimodal biologic data including at least one of a gene expression level, mRNA, metabolic reaction fluxes, or intracellular metabolite concentrations from biologic systems(Kumaran, pg., 29, see also figs. 2-5, “As shown in Figure 3, G-2 (as described in Figure 2) produces key extracellular metabolites (DOX+ME+MEcPP) equivalent to 3.45 g/L of terpenoid product. This compares to 7 g/L of primarily DOX in the strain not containing modifications to increase Fe-S proteins. G4 also had a 56% higher accumulation of the terpenoid
product. Genetic modifications include
∆
ryhB,
∆
iscR with overexpression of the isc operon[collecting multimodal biologic data including at least one of a gene expression level metabolic reaction fluxes, or intracellular metabolite concentrations from biologic systems].”). 1
processing the collected biologic data [through data normalization] and quality assurance steps to create model-ready data(Kumaran, pgs., 30-31, see also figs. 13-15, “At the endpoint, the cell cultures are measured for OD600, then are split into two samples; the first is extracted with Ethyl Acetate to analyze and quantify Product A, while the second is centrifuged to pellet the cells, and the supernatant is analyzed for extracellular MEP metabolites while the cell pellet is further processed to extract intracellular MEP metabolites[processing the collected biologic data]. The terpenoid product-containing organic phase sample is analyzed via gas chromatography and mass spectrometry (GC/MS), while the extracellular and intracellular metabolite samples are separated and analyzed via liquid chromatography and mass spectrometry (LC/NlS). The LC/MS detector is a triple-quadrupole instrument, enabling accurate quantification of MEP metabolites against authentic standards[and quality assurance steps to create model-ready data]”).2
and generating [by the trained machine learning model] at least one output predicting an effect of genetic modification on a metabolite level or a reaction flux(Kumaran, pgs., 30-31, see also figs. 15, “As shown in Figure 15, metabolomics analysis of Product A GS strains with MEP complementation with dxr or dxr-ispE further shows that the increased MEP flux potential in the resulting strains translates into more potential Product A titer. The measured product and MEP metabolite concentrations for these strain cultures are converted to molarity, and the carbon equivalent of each MEP metabolite in terms of Product A is calculated[and generating at least one output predicting an effect of genetic modification on a metabolite level or a reaction flux]... [w]ith balanced expression between MEP genes, up to 12x more product A could result.” ).3, 4
Kumaran does not teach: through data normalization
However, Volk teaches:
through data normalization(Volk, pg., 9, see also fig. 4, “At first, a library of random variants was generated, and the activities of its members were measured. The sequences of
these variants were encoded as one-hot vectors, meaning each position on the sequence was represented by a vector with a “1” at the index of the amino acid present and a “0” at all other
indices[through data normalization].”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran with the teachings of Volk the motivation to do so would be to incorporate machine learning during the bioengineering process to engineer better solutions(Volk, pgs., 1-2, “[D]ue to the complexity of biological systems, performing biosystems design in a quantitative and predictive manner remains an overwhelming
Challenge...[t]o identify patterns and trends in the data-rich efforts produced by systems biology, machine learning (ML) has surfaced as a potential solution to hasten success in biosystems design.”).
Kumaran in view of Volk does not teach: wherein the data normalization includes applying a batch effect correction to minimize batch-specific systemic variation across experimental runs, wherein the batch-specific systemic variation across experimental runs arises from one or more of: laboratory equipment variations, plate-to-plate variations, or plate lot variations; training a machine learning model using the model-ready data; by the trained machine learning model.
However, Stajdohar teaches:
wherein the data normalization includes applying a batch effect correction to minimize batch-specific systemic variation across experimental runs(Stajdohar, paras., [0063-0064], see also fig. 8., “A test dataset 801 may consist of measurements from one or more patients, patient samples, or experimental specimens (for example, gene expression values from a cohort of patients in a clinical trial)[ batch-specific systemic variation across experimental runs]…[t]he preprocessor function is selected from the library of preprocessors 803… [c]orrectively, the library of preprocessors 803 and classifier 804 are referred to as signature model 805[wherein the data normalization includes applying a batch effect correction to minimize].”),
wherein the batch-specific systemic variation across experimental runs arises from one or more of: laboratory equipment variations, plate-to-plate variations, or plate lot variations; training a machine learning model using the model-ready data(Stajdohar, paras., [0087-0088], “[A] Sample Handling preprocessor is provided. A Sample Handling Preprocessor is trained to mitigate the batch effect introduced when tissue samples are handled differently prior to data generation. Bias modalities may be associated with any number of tissue collection, storage and processing methods. In the cancer diagnostics space, for example, samples may come from fresh frozen biopsy, biopsy aspirate, core needle biopsy, formalin fixed paraffin embedded (FFPE) slides, etc[wherein the batch-specific systemic variation across experimental runs arises from one or more of: laboratory equipment variations]. The duration of time the samples have been stored, e.g., fresh vs. archival, introduces bias. Further, the manner in which tissue has been collected for extraction and sequencing, such as microdissection vs scrape all, introduces bias. Such a preprocessor is useful where a model is developed for use in a clinical trial assay (CTA), such as the Xema TME Panel[training a machine learning model using the model-ready data]… [a] Sample Handling preprocessor would enable the trained model to be deployed for both the analytical validation and clinical trial use.”);5
by the trained machine learning model(Stajdohar, paras., [0087-0088], “[A] preprocessor is useful where a model is developed for use in a clinical trial assay (CTA), such as the Xema TME Panel[machine learning model]…[a] Sample Handling preprocessor would enable the trained model[by the trained] to be deployed for both the analytical validation and clinical trial use.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk with the teachings of Stajdohar the motivation to do so would be to develop a generalized machine learning model that can accurately make predictions even though batch effects are present in the dataset(Stajdohar, paras. [0030-0032], “[C]omputational models often fail on new patients or biological samples due to technical bias introduced by assay type or sample handling. As with disease type, one must account for sources of technical bias… [i]n particular, there is a need for methods enabling a single predictive model, already trained and with the algorithm locked, to work across different disease areas, patient groups, technical assay types, etc.”).
Regarding claim 2, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, wherein the biologic data is converted from a first structured format to a second format suitable for model training(Volk, pg., 9, see also fig. 4, “At first, a library of random variants was generated, and the activities of its members were measured[wherein the normalized biologic data is converted from a first structured format]. The sequences of
these variants were encoded as one-hot vectors, meaning each position on the sequence was represented by a vector with a “1” at the index of the amino acid present and a “0” at all other
indices[to a second format suitable for model training].”).6
Regarding claim 3, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, wherein training the machine learning model using the model-ready data comprises: training the machine learning model to process a model input comprising genetic perturbation data to generate a model output that predicts a cellular phenotype associated with the genetic perturbation defined by the genetic perturbation data (Volk, pg., 12, see also fig. 7, “[T]o further characterize the metabolism-regulatory mechanism by observing metabolome and proteome changes in 97 kinase S. cerevisiae mutants (Figure 7)...ML...can be used to map changes in regular enzyme expression levels, which could then be used to predict the metabolic phenotype”).7 8
Regarding claim 4, Kumaran in view of Volk and Stajdohar teaches the method of claim 3, wherein training the machine learning model comprises: using a knowledge graph to represent biological entities as nodes; representing relationships between entities as edges; and capturing biological relationships in a format appropriate for processing by the machine learning model(Volk, pg., 12, As fig. 7 details below:
PNG
media_image1.png
271
768
media_image1.png
Greyscale
In step (b) a graph topology was created where the nodes represented the metabolites and enzymes and the edges represented the strength of the regulation between metabolites and enzymes during the metabolic process. This graph representation was used to construct a multiple linear regression model (MLR) to predict metabolite concentrations for mutated organisms of the S. cerevisiae species).
Regarding claim 5, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, wherein collecting multimodal biological data comprises: obtaining RNA sequencing data for genome-wide gene expression levels; measuring metabolic reaction fluxes; and collecting metabolite concentration data using mass spectrometry(Kumaran, pgs., 30-31, see also figs. 11-15, “Decreasing the amount of ispB enzyme in the cell by modifying the ispB RBS sequence and tuning down translation of the mRNA could shift carbon flux toward the recombinant terpenoid pathway[obtaining RNA sequencing data for genome-wide gene expression levels]... [t]he LC/MS detector is a triple-quadrupole instrument, enabling accurate quantification of MEP metabolites[and collecting metabolite concentration data using mass spectrometry]...[a]s shown in Figure 15, metabolomics analysis of Product A GS strains with MEP complementation with dxr or dxr-ispE further shows that the increased MEP flux potential in the resulting strains translates into more potential Product A titer[measuring metabolic reaction fluxes].”).
Regarding claim 6, Kumaran in view of Volk and Stajdohar teaches the method of claim 5, wherein the mass spectrometry is liquid chromatography–mass spectrometry(Kumaran, pgs., 30-31, see also figs. 11-15, “[T]he extracellular and intracellular metabolite samples are separated and analyzed via liquid chromatography and mass spectrometry (LC/MS)[wherein the mass spectrometry is liquid chromatography–mass spectrometry].”).
Regarding claim 7, Kumaran in view of Volk and Stajdohar teaches the method of claim 5, wherein the mass spectrometry is gas chromatography–mass spectrometry(Kumaran, pgs., 30-31, see also figs. 11-15, “The terpenoid product-containing organic phase sample is analyzed via gas chromatography and mass spectrometry (GC/MS)[ wherein the mass spectrometry is gas chromatography–mass spectrometry]....”).
Regarding claim 14, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, wherein collecting the multimodal biologic data further comprises:
combining gene expression data from RNA sequencing; incorporating flux data from an isotope-labeled experiment(Kumaran, pgs., 27-28, “Further, to improve MEP carbon, the E coli has one or more of the following genetic modifications... decreased expression or activity of IspB... ryhB deletion or inactivation[combining gene expression data from RNA sequencing]... a mutation in pgi (glucose-6-phosphate isomerase) supporting increased terpenoid titer or increases in MEP carbon[incorporating flux data from an isotope-labeled experiment]... the bacterial strains comprises...all 6 modifications defined by (1) to (6) above.”);
and merging a metabolite concentration measurement from mass spectrometry(Kumaran, pgs., 30-31, see also figs. 13-15, “At the endpoint, the cell cultures are measured for OD600, then are split into two samples; the first is extracted with Ethyl Acetate to analyze and quantify Product A, while the second is centrifuged to pellet the cells, and the supernatant is analyzed for extracellular MEP metabolites while the cell pellet is further processed to extract intracellular MEP metabolites. The terpenoid product-containing organic phase sample is analyzed via gas chromatography and mass spectrometry (GC/MS), while the extracellular and intracellular metabolite samples are separated and analyzed via liquid chromatography and mass spectrometry (LC/NlS)[ and merging a metabolite concentration measurement from mass spectrometry].”).
Regarding claim 16, Kumaran in view of Volk, and Stajdohar teaches the method of claim 1, wherein the multimodal biologic data derives from at least one integrated sensor(Volk, pg., 12, see also fig., 8, “Automated monitoring of meaningful variables during the
fermentation process is often impossible, but soft sensors allow for correlation between easily measured online and offline variables to real time prediction of meaningful offline variables[wherein the multimodal biologic data derives from at least one integrated sensor].”).9
Regarding claim 20, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, wherein training the machine learning model using the model-ready data comprises: training the machine learning model to predict effects of genetic modifications on metabolic pathways; training the machine learning model to predict changes in metabolite concentrations; or training the machine learning model to predict reaction flux distributions in response to genetic perturbations(Kumaran, pgs., 30-31, see also figs. 15, “As shown in Figure 15, metabolomics analysis of Product A GS strains with MEP complementation with dxr or dxr-ispE further shows that the increased MEP flux potential in the resulting strains translates into more potential Product A titer[training the machine learning model to predict effects of genetic modifications on metabolic pathways; or training the machine learning model to predict reaction flux distributions in response to genetic perturbations]. The measured product and MEP metabolite concentrations for these strain cultures are converted to molarity, and the carbon equivalent of each MEP metabolite in terms of Product A is calculated... [w]ith balanced expression between MEP genes, up to 12x more product A could result[training the machine learning model to predict changes in metabolite concentrations].”).
Regarding claim 21, Stajdohar teaches a system comprising: one or more computers; and one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations(Stajdohar, paras., [0101-0107], see also fig. 15, “As shown in FIG. 15, computer system/server 12 in computing node 10 is shown in the form of a general purpose computing device. The components of computer system/server 12 may include, but are not limited to, one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including system memory 28 to processor 16… [t]he computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present disclosure.”) and for all other claim limitations they are rejected on the same basis as independent claim 1 since they are analogous claims.
Regarding claim 22, Stajdohar teaches one or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations(Stajdohar, paras., [0101-0107], see also fig. 15, “As shown in FIG. 15, computer system/server 12 in computing node 10 is shown in the form of a general purpose computing device. The components of computer system/server 12 may include, but are not limited to, one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including system memory 28 to processor 16… [t]he computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present disclosure.”) and for all other claim limitations they are rejected on the same basis as independent claim 1 since they are analogous claims.
Regarding dependent claims 23-24, they are rejected on the same basis as dependent claims 2-3 since they are analogous claims.
Claims 10-11 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Kumaran et al., CA 3051463-A1(“Kumaran”) in view of Volk, Michael Jeffrey, et al. "Biosystems design by machine learning." ACS synthetic biology 9.7 (2020)(“Volk”) and in view of Stajdohar et al., US 2024/0233874 Al(“Stajdohar”) and further in view of Otte et al., US 11,514,289 Bl(“Otte”)
Regarding claim 10, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, but do not teach: wherein at least some of the multimodal biologic data is collected using a rapid sampling system.
However Otte teaches:
wherein at least some of the multimodal biologic data is collected using a rapid sampling system(Otte, cols. 8-9, lines 45-67 & lines 1-21, “[A] sample is taken at a first time point and sequenced, and then another sample is taken at a subsequent time point and is sequenced. Thus, two samples can be collected from a subject at different times, e.g., one being at a time when the subject was healthy or had a different biological condition. One of the samples can be treated as a normal control or reference sample, from which differences of a sample can be identified.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk and Stajdohar with the teachings of Otte the motivation to do so would be to devise a system in which the variables associated with genetic data can be transformed into usable formats for training machine learning and/or artificial intelligence models(Otte, cols. 1-2, lines 15-52, “[T]he amount of variables that can potentially be used as inputs to a model can be voluminous, as can be encountered with genetic data. Training an ML model can be difficult using a large amount of input variables. And, knowing which input variables to use and how to structure the input variables is not straightforward... [a] set of input features for training the machine learning model can be identified and used to train the model based on training samples... the input features can include aligned variables ( e.g., derived from DNA sequences aligned to a population level or individual reference genome) and/or non-aligned variables (e.g., sequence content).”).
Regarding claim 11, Kumaran in view of Volk, Stajdohar, and Otte teaches the method of claim 10, wherein the rapid sampling system comprises:
automated sampling mechanisms for collecting standardized samples; near-instantaneous quenching of cellular metabolism(Volk, pg., 11, see also fig. 6, “ML in pathway engineering...developed a platform called BioAutomata, which optimized the lycopene biosynthetic pathway in E. coli by changing the promoter strength of three promoters governing the transcription of the target pathway[near-instantaneous quenching of cellular metabolism]. A DBTL cycle was implemented to construct and test variants and learn from the acquired data to modify promoter strength for the next iteration of the cycle. Using a Bayesian optimization scheme that was first validated on a simple two dimensional function, the automated system evaluated less than 1% of the possible variants and outperformed a random screen, analogous to a high throughput screen, by 77%[ automated sampling mechanisms for collecting standardized samples].”);
and integration with liquid chromatography–mass spectrometry and gas chromatography–mass spectrometry for metabolite analysis(Kumaran, pgs., 30-31, see also figs. 13-15, “The terpenoid product-containing organic phase sample is analyzed via gas chromatography and mass spectrometry (GC/MS), while the extracellular and intracellular metabolite samples are separated and analyzed via liquid chromatography and mass spectrometry (LC/NlS). The LC/MS detector is a triple-quadrupole instrument, enabling accurate quantification of MEP metabolites against authentic standards[and integration with liquid chromatography–mass spectrometry and gas chromatography–mass spectrometry for metabolite analysis]....”).10
Regarding claim 13, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, but do not teach further comprising: tracking data lineage from a raw experimental measurement to a processed value; maintaining metadata about experimental conditions; and validating a normalization procedure using a control sample.
However, Otte teaches:
further comprising: tracking data lineage from a raw experimental measurement to a processed value; maintaining metadata about experimental conditions (Otte, cols. 11-12, see also fig. 3, “FIG. 3 shows a system 300 for generating a machine learning model using genetic data... [i]nstrument 305 may include any physical components including hardware, software, and biochemical products, which may be used to measure genetic data 306 from a sample of a subject... [n]on-genetic data 307 can include, but not limited to, age, weight, height, ethnicity, and other such physical characteristics, as well as names, dates of birth, genders, demographics, measurements of mental, physical, or physiological wellness, medical history, sample sources, sample collection times, and sample biological conditions.”);
and validating a normalization procedure using a control sample (Otte, col. 8-9, lines 60-67 and lines 1-21, “[T]wo samples can be collected from a subject at different times, e.g., one
being at a time when the subject was healthy or had a different biological condition. One of the samples can be treated as a normal control or reference sample, from which differences of a sample can be identified.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk and Stajdohar with the teachings of Otte the motivation to do so would be to devise a system in which the variables associated with genetic data can be transformed into usable formats for training machine learning and/or artificial intelligence models(Otte, cols. 1-2, lines 15-52, “[T]he amount of variables that can potentially be used as inputs to a model can be voluminous, as can be encountered with genetic data. Training an ML model can be difficult using a large amount of input variables. And, knowing which input variables to use and how to structure the input variables is not straightforward... [a] set of input features for training the machine learning model can be identified and used to train the model based on training samples... the input features can include aligned variables ( e.g., derived from DNA sequences aligned to a population level or individual reference genome) and/or non-aligned variables (e.g., sequence content).”).
Claims 15 and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Kumaran et al., CA 3051463-A1(“Kumaran”) in view of Volk, Michael Jeffrey, et al. "Biosystems design by machine learning." ACS synthetic biology 9.7 (2020)(“Volk”) and in view of Stajdohar et al., US 2024/0233874 Al(“Stajdohar”) and further in view of Tang, Xin, et al. "Explainable multi-task learning for multi-modality biological data analysis." Nature communications 14.1 (2023)(“Tang”)
Regarding claim 15, Kumaran in view of Volk and Stajdohar teaches the method of claim 1 but do not teach: wherein the machine learning model comprises a neural network with separate encoding branches for different data modalities; and wherein generating, by the trained machine learning model, at least one output predicting an effect of genetic modification on a metabolite level or a reaction flux comprises: processing the multimodal biologic data using the separate encoding branches to generate a respective encoded representation for each data modality; combining the encoded representations of the data modalities through fusion layers of the neural network; and processing an output of the fusion layers by an output layer of the neural network to generate the at least one output.
However, Tang teaches:
wherein the machine learning model comprises a neural network with separate encoding branches for different data modalities(Tang, pgs., 12-17, see also fig. 1, “For each modality v=1,...,V, UnitedNet[a neural network] has one encoder
E
n
c
v
(
⋅
)
that maps the features of each cell i (i=1,...,n) to a modality-specific latent code
z
i
(
v
)
containing the most essential information of the data:
z
i
(
v
)
=
E
n
c
v
(
x
i
v
)
[ with separate encoding branches for different data modalities]….”);
and wherein generating, by the trained machine learning model, at least one output predicting an effect of genetic modification on a metabolite level or a reaction flux comprises:
processing the multimodal biologic data using the separate encoding branches to generate a respective encoded representation for each data modality(Tang, pgs., 12-17, see also fig. 1, “Let
X
(
v
)
∈
R
n
×
p
v
v
=
1
,
…
,
V
denote the data from modality v where its i th row
x
i
(
v
)
… [f]or each modality v=1, . . . ,V, UnitedNet has one encoder
E
n
c
v
(
⋅
)
that maps the features of each cell i(i=1,…,n) to a modality-specific latent code….”);
combining the encoded representations of the data modalities through fusion layers of the neural network(Tang, pgs., 12-17, see also fig. 1, “Denote the number of groups to be K. The group identification module takes the modality-specific codes from all modalities (
z
i
(
1
)
,
…
,
z
i
(
V
)
) as the input and assigns it to one of the K groups. It first fuses the data:
z
i
=
∑
v
=
1
V
η
v
z
i
(
v
)
”);
and processing an output of the fusion layers by an output layer of the neural network to generate the at least one output(Tang, pgs., 12-17, see also fig. 1, “Next, the fused representation
z
i
is passed through a fully connected layer... the intermediate output
h
i
is processed by another fully connected layer…[t]he group identification module assigns the group index
k
i
=
a
r
g
m
a
x
k
=
1
,
…
,
K
M
k
(
h
i
)
to cell i. ”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk and Stajdohar with the teachings of Tang the motivation to do so would be to use one model to predict multiple biochemistry metrics when working with multi-modal biological data(Tang, pgs., 1-2, “Recent advances in single-cell biotechnology make it possible to simultaneously measure gene expression along with other high-dimensional modalities for the same cells... [c]ompared with single-modality analysis, recent studies have identified more tasks for multi-modality analysis7 such as (i) identification of biologically meaningful groups from different modalities, enabling a deeper biological understanding of cellular identities and functions for different biological systems, and (ii) cross-modal prediction among different modalities, inferring the information of cells that cannot be easily or simultaneously measured.”).
Regarding claim 18, Kumaran in view of Volk, Stajdohar, and Tang teaches the method of claim 15, wherein the neural network comprises: separate encoding branches for gene expression data(Tang, pgs., 6-7, see also fig. 1, “[T]he single-cell transposase-accessible DNA sequencing technique5 that combines gene expression and genome wide DNA accessibility (namely, multiome ATAC + gene expression data) has been used to profile diverse types of immune cells... UnitedNet...can perform a supervised group identification task (termed annotation transfer) that automatically identifies cell types in new, unlabeled test samples based on previously labeled training samples, while simultaneously enabling cross-modal prediction[the separate encoding branches for gene expression data].”); 11dedicated pathways for metabolite profile processing(Volk, pgs., 11-12, see also fig. 7, “[T]he metabolism-regulatory mechanism by observing metabolome and proteome changes in 97 kinase S. cerevisiae mutants (Figure 7)...can be used to map changes in regular enzyme expression levels, which could then be used to predict the metabolic phenotype[dedicated pathways for metabolite profile processing].”); and specialized branches for reaction flux analysis(Volk, pgs., 11-12, see also fig. 7, “[S]upervised ML methods and flux balance analysis (FBA) have been used together to predict bacterial central metabolism by using input features from 37 bacteria species which all have C13 metabolic flux data[and specialized branches for reaction flux analysis].”).12
Regarding claim 19, Kumaran in view of Volk and Stajdohar teaches the method of claim 1, but do not teach: further comprising: normalizing the multimodal biologic data across different organisms and conditions.
However, Tang teaches:
normalizing the multimodal biologic data across different organisms and conditions(Tang, pg., 17, “We use the DBiT-seq embryo dataset, where the following three modalities of 936 spots in DBiT-seq are taken: mRNA expression, protein expression, and niche mRNA expression. For modality of mRNA expression, we normalize the raw count matrix using function scanpy.pp.normalize_total from scanpy and select the top 568 differentially expressed genes. For the modality of protein expression, we normalize the raw count matrix and used 22 kinds of proteins. The niche modalities are generated based on normalized mRNA expression[normalizing the multimodal biologic data across different organisms and conditions].”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk and Stajdohar with the teachings of Tang the motivation to do so would be to use one model to predict multiple biochemistry metrics when working with multi-modal biological data(Tang, pgs., 1-2, “Recent advances in single-cell biotechnology make it possible to simultaneously measure gene expression along with other high-dimensional modalities for the same cells... [c]ompared with single-modality analysis, recent studies have identified more tasks for multi-modality analysis7 such as (i) identification of biologically meaningful groups from different modalities, enabling a deeper biological understanding of cellular identities and functions for different biological systems, and (ii) cross-modal prediction among different modalities, inferring the information of cells that cannot be easily or simultaneously measured.”).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ADAM C STANDKE whose telephone number is (571)270-1806. The examiner can normally be reached Gen. M-F 9-9PM EST.
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, Michael J Huntley can be reached at (303) 297-4307. 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.
/Adam C Standke/
Primary Examiner
Art Unit 2129
1 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
2 Examiner Notes: The claim limitations that are not in bold and contained within square brackets (i.e., [ ]) are
claim limitations that are not taught by the prior art of Kumaran.
3 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
4 Examiner Notes: The claim limitations that are not in bold and contained within square brackets (i.e., [ ]) are
claim limitations that are not taught by the prior art of Kumaran.
5 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
6 It would have been obvious to one of ordinary skill in the art before the effective filing date
of the claimed invention to modify the teachings of Kumaran with the above teachings of Volk for the same rationale stated at Claim 1.
7 According to the broadest reasonable interpretation (BRI), the use of alternative language amounts to the claim requiring one or more elements but not all.
8 It would have been obvious to one of ordinary skill in the art before the effective filing date
of the claimed invention to modify the teachings of Kumaran with the above teachings of Volk for the same rationale stated at Claim 1.
9 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran with the above teachings of Volk for the same rationale stated at Claim 1.
10 It would have been obvious to one of ordinary skill in the art before the effective filing date
of the claimed invention to modify the teachings of Kumaran with the above teachings of Volk for the same rationale stated at Claim 1.
11 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran in view of Volk and Stajdohar with the above teachings of Tang for the same rationale stated at Claim 15.
12 It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Kumaran with the above teachings of Volk for the same rationale stated at Claim 1.
Prosecution Insights