Systems and methods for diagnosing a disease or a condition

A computer system for constructing a model that determines whether a subject is afflicted with a condition, the computer system comprising:

• one or more processors;
• and
• memory addressable by the one or more processors, the memory storing at least one program for execution by the one or more processors, the at least one program comprising instructions for:
• A) for each respective first subject in a first plurality of subjects not afflicted with the condition,
• obtaining a first RNA-seq dataset comprising a respective discrete attribute value for each gene transcript in a corresponding first plurality of gene transcripts, for each cell in a respective first plurality of cells from a corresponding first biological sample from the respective first subject, and
• obtaining a first ATAC-seq dataset comprising a respective ATAC fragment count for each corresponding ATAC peak in a corresponding first plurality of ATAC peaks, for each respective cell in a respective second plurality of cells from a corresponding second biological sample from the respective subject;
• B) for each respective second subject in a second plurality of subjects afflicted with the condition,
• obtaining a second RNA-seq dataset comprising a respective discrete attribute value for each gene transcript in a corresponding second plurality of gene transcripts, for each cell in a respective third plurality of cells from a corresponding third biological sample from the respective second subject, and
• obtaining a second ATAC-seq dataset comprising a respective ATAC fragment count for each ATAC peak in a corresponding second plurality of ATAC peaks, for each respective cell in a respective fourth plurality of cells from a corresponding fourth biological sample from the respective subject;
• C) using the first RNA-seq dataset and the second RNA-seq dataset to identify a plurality of candidate genes having differential transcription;
• D) using the first ATAC-seq dataset and the second ATAC-seq dataset to identify a plurality of candidate ATAC peaks having differential accessibility between the first plurality of subjects and the second plurality of subjects;
• E) for each respective transcription factor motif in a plurality of transcription factor motifs, mapping the respective transcription factor motif onto the plurality of candidate ATAC peaks form a plurality of mapped transcription factor motifs;
• and
• F) constructing the model that determines whether a subject is afflicted with a condition using ATAC-seq abundance data in the first and second RNA-seq dataset for those candidate genes in the plurality of candidate genes satisfying a proximity threshold with respect to a respective candidate ATAC peak to which a transcription factor motif in the plurality of transcription factor motifs mapped.

A non-transitory computer readable storage medium, wherein the non-transitory computer readable storage medium stores instructions, which when executed by a computer system, cause the computer system to perform a method for constructing a model that determines whether a subject is afflicted with a condition, the method comprising:

• A) for each respective first subject in a first plurality of subjects not afflicted with the condition,
• obtaining a first RNA-seq dataset comprising a respective discrete attribute value for each gene transcript in a corresponding first plurality of gene transcripts, for each cell in a respective first plurality of cells from a corresponding first biological sample from the respective first subject, and
• obtaining a first ATAC-seq dataset comprising a respective ATAC fragment count for each corresponding ATAC peak in a corresponding first plurality of ATAC peaks, for each respective cell in a respective second plurality of cells from a corresponding second biological sample from the respective subject;
• B) for each respective second subject in a second plurality of subjects afflicted with the condition,
• obtaining a second RNA-seq dataset comprising a respective discrete attribute value for each gene transcript in a corresponding second plurality of gene transcripts, for each cell in a respective third plurality of cells from a corresponding third biological sample from the respective second subject, and
• obtaining a second ATAC-seq dataset comprising a respective ATAC fragment count for each ATAC peak in a corresponding second plurality of ATAC peaks, for each respective cell in a respective fourth plurality of cells from a corresponding fourth biological sample from the respective subject;
• C) using the first RNA-seq dataset and the second RNA-seq dataset to identify a plurality of candidate genes having differential transcription;
• D) using the first ATAC-seq dataset and the second ATAC-seq dataset to identify a plurality of candidate ATAC peaks having differential accessibility between the first plurality of subjects and the second plurality of subjects;
• E) for each respective transcription factor motif in a plurality of transcription factor motifs, mapping the respective transcription factor motif onto the plurality of candidate ATAC peaks form a plurality of mapped transcription factor motifs;
• and
• F) constructing the model that determines whether a subject is afflicted with a condition using ATAC-seq abundance data in the first and second RNA-seq dataset for those candidate genes in the plurality of candidate genes satisfying a proximity threshold with respect to a respective candidate ATAC peak to which a transcription factor motif in the plurality of transcription factor motifs mapped.

A method of evaluating a gene signature associated with a target condition that can afflict a host species, wherein the gene signature comprises a first plurality of positive genes that are up-regulated when the test subject has the target condition and a second plurality of genes that are down-regulated when the test subject has the target condition, the method comprising:

• A) obtaining an indication of each gene in the first plurality of positive genes;
• B) obtaining an indication of each gene in the second plurality of negative genes;
• C) obtaining a plurality of datasets, wherein
• each dataset in the plurality of datasets includes transcriptional data for each respective subject in a corresponding plurality of subjects and an indication of whether the respective subject has or does not have a respective test condition in a plurality of test conditions,
• the plurality of datasets includes at least one dataset for each test condition in the plurality of test conditions,
• at least one test condition in the plurality of test conditions is the target condition,
• D) for each respective dataset in a plurality of datasets, for each respective time point in a set of time points represented by the respective dataset:
• for each respective subject in the respective dataset, determining a score for the respective subject at the respective time point by determining a difference between a geometric mean of abundance values for the first plurality of positive genes and a geometric mean of abundance values for the second plurality of positive genes indicated in the respective dataset,
• determining an area under a receiver operator characteristic curve (AUROC) value for the respective dataset for the test condition using the respective score for each subject in the respective dataset at each respective timepoint;
• E) evaluating a performance of the gene signature using the AUROC value of each dataset in the plurality of datasets associated with the target condition;
• and
• F) evaluating a cross-reactivity of the gene signature from the AUROC value of each dataset in the plurality of datasets associated with a test condition that is other than the target condition.

A computer system for evaluating a gene signature associated with a target condition that can afflict a host species, wherein the gene signature comprises a first plurality of positive genes that are up-regulated when the test subject has the target condition and a second plurality of genes that are down-regulated when the test subject has the target condition, the computer system comprising:

• one or more processors;
• and
• memory addressable by the one or more processors, the memory storing at least one program for execution by the one or more processors, the at least one program comprising instructions for:
• A) obtaining an indication of each gene in the first plurality of positive genes;
• B) obtaining an indication of each gene in the second plurality of negative genes;
• C) obtaining a plurality of datasets, wherein
• each dataset in the plurality of datasets includes transcriptional data for each respective subject in a corresponding plurality of subjects and an indication of whether the respective subject has or does not have a respective test condition in a plurality of test conditions,
• the plurality of datasets includes at least one dataset for each condition in the plurality of test conditions,
• at least one test condition in the plurality of test conditions is the target condition,
• D) for each respective dataset in a plurality of datasets, for each respective time point in a set of time points represented by the respective dataset:
• for each respective subject in the respective dataset, determining a score for the respective subject at the respective time point by determining a difference between a geometric mean of abundance values for the first plurality of positive genes and a geometric mean of abundance values for the second plurality of positive genes indicated in the respective dataset,
• determining an area under a receiver operator characteristic curve (AUROC) value for the respective dataset for the test condition using the respective score for each subject in the respective dataset at each respective timepoint;
• E) evaluating a performance of the gene signature using the AUTROC value of each dataset in the plurality of datasets associated with the target condition;
• and
• F) evaluating a cross-reactivity of the gene signature from the AUROC value of each dataset in the plurality of datasets associated with a test condition that is other than the target condition.

A non-transitory computer readable storage medium, wherein the non-transitory computer readable storage medium stores instructions, which when executed by a computer system, cause the computer system to perform a method for evaluating a gene signature associated with a target condition that can afflict a host species, wherein the gene signature comprises a first plurality of positive genes that are up-regulated when the test subject has the target condition and a second plurality of genes that are down-regulated when the test subject has the target condition, the method comprising:

• A) obtaining an indication of each gene in the first plurality of positive genes;
• B) obtaining an indication of each gene in the second plurality of negative genes;
• C) obtaining a plurality of datasets, wherein
• each dataset in the plurality of datasets includes transcriptional data for each respective subject in a corresponding plurality of subjects and an indication of whether the respective subject has or does not have a respective test condition in a plurality of test conditions,
• the plurality of datasets includes at least one dataset for each condition in the plurality of test conditions,
• at least one test condition in the plurality of test conditions is the target condition,
• D) for each respective dataset in a plurality of datasets, for each respective time point in a set of time points represented by the respective dataset:
• for each respective subject in the respective dataset, determining a score for the respective subject at the respective time point by determining a difference between a geometric mean of abundance values for the first plurality of positive genes and a geometric mean of abundance values for the second plurality of positive genes indicated in the respective dataset,
• determining an area under a receiver operator characteristic curve (AUROC) value for the respective dataset for the test condition using the respective score for each subject in the respective dataset at each respective timepoint;
• E) evaluating a performance of the gene signature using the AUROC value of each dataset in the plurality of datasets associated with the target condition;
• and
• F) evaluating a cross-reactivity of the gene signature from the AUROC value of each dataset in the plurality of datasets associated with a test condition that is other than the target condition.

A computer system for determining whether a subject is infected with SARS-CoV-2, the computer system comprising:

• one or more processors;
• and
• memory addressable by the one or more processors, the memory storing at least one program for execution by the one or more processors, the at least one program comprising instructions for:
• obtaining, in electronic form, a plurality of discrete attribute values, wherein each discrete attribute value in the plurality of discrete attribute values represents a transcript abundance of a respective gene in a plurality of genes in a biological sample from the subject, wherein the plurality of genes comprises three or more genes in the group consisting of PIF1, BANF1, ROCK2, DOCK5, SLK, TVP23B, GUDC1, ARAP2, SLC25A46, TCEAL3, and EHD3;
• and
• inputting the plurality of discrete attribute values into a model comprising a plurality of parameters, wherein the model applies the plurality of parameters to the plurality of discrete attribute values to generate as output from the model an indication as to whether the subject is infected with SARS-CoV-2.

A method for determining whether a subject has a characteristic, the method comprising:

• sequencing a plurality of mRNA molecules from a biological sample obtained from the subject, thereby obtaining a plurality of sequence reads of RNA from the subject;
• aligning each respective sequence read in the plurality of sequence reads to a reference human transcriptome, thereby obtaining a corresponding plurality of aligned sequence reads;
• using the corresponding plurality of aligned sequence reads to determine a corresponding transcript abundance in a plurality of transcript abundances, wherein each respective transcript abundance in the plurality of transcript abundances represents a transcript abundance of a corresponding gene in a plurality of genes;
• inputting the plurality of transcript abundances into each respective neural network in a plurality of neural networks, wherein each respective neural network in the plurality of neural networks represents a different gene set in a plurality of gene sets, and wherein each respective neural network in the plurality of neural networks comprises:
• (a) a corresponding plurality of input nodes, each respective input node in the corresponding plurality of input nodes for a different transcript abundance in the plurality of transcript abundance abundances, and
• (b) a representation of the corresponding gene set in the form of (i) a corresponding plurality of hidden nodes, each hidden node representing a gene in the corresponding gene set, and (ii) a corresponding plurality of edges, wherein each edge in the corresponding plurality of edges interconnects an input node in the plurality of input nodes to a hidden node in the corresponding plurality of hidden nodes with a corresponding edge weight;
• responsive to the inputting, obtaining a plurality of predictions, each prediction in the plurality of predictions from a neural network in the plurality of neural networks;
• and
• responsive to inputting the plurality of predictions into an ensemble model obtaining, as output form the ensemble model a prediction of whether the subject has the characteristic.

A computer system for determining whether a subject has a characteristic, the computer system comprising:

• one or more processors;
• and
• memory addressable by the one or more processors, the memory storing at least one program for execution by the one or more processors, the at least one program comprising instructions for:
• aligning each respective sequence read in a plurality of sequence reads, wherein the plurality of sequence reads represent a plurality of mRNA molecules in a biological sample obtained from the subject, to a reference human transcriptome, thereby obtaining a corresponding plurality of aligned sequence reads;
• using the corresponding plurality of aligned sequence reads to determine a corresponding transcript abundance in a plurality of transcript abundances, wherein each respective transcript abundance in the plurality of transcript abundances represents a transcript abundance of a corresponding gene in a plurality of genes;
• inputting the plurality of transcript abundances into each respective neural network in a plurality of neural networks, wherein each respective neural network in the plurality of neural networks represents a different gene set in a plurality of gene sets, and wherein each respective neural network in the plurality of neural networks comprises:
• (a) a corresponding plurality of input nodes, each respective input node in the corresponding plurality of input nodes for a different transcript abundance in the plurality of transcript abundance abundances, and
• (b) a representation of the corresponding gene set in the form of (i) a corresponding plurality of hidden nodes, each hidden node representing a gene in the corresponding gene set, and (ii) a corresponding plurality of edges, wherein each edge in the corresponding plurality of edges interconnects an input node in the plurality of input nodes to a hidden node in the corresponding plurality of hidden nodes with a corresponding edge weight;
• responsive to the inputting, obtaining a plurality of predictions, each prediction in the plurality of predictions from a neural network in the plurality of neural networks;
• and
• responsive to inputting the plurality of predictions into an ensemble model obtaining, as output form the ensemble model a prediction of whether the subject has the characteristic.

A non-transitory computer readable storage medium, wherein the non-transitory computer readable storage medium stores instructions, which when executed by a computer system, cause the computer system to perform a method for determining whether a subject has a characteristic, the method comprising:

• aligning each respective sequence read in a plurality of sequence reads, wherein the plurality of sequence reads represent a plurality of mRNA molecules in a biological sample obtained from the subject, to a reference human transcriptome, thereby obtaining a corresponding plurality of aligned sequence reads;
• using the corresponding plurality of aligned sequence reads to determine a corresponding transcript abundance in a plurality of transcript abundances, wherein each respective transcript abundance in the plurality of transcript abundances represents a transcript abundance of a corresponding gene in a plurality of genes;
• inputting the plurality of transcript abundances into each respective neural network in a plurality of neural networks, wherein each respective neural network in the plurality of neural networks represents a different gene set in a plurality of gene sets, and wherein each respective neural network in the plurality of neural networks comprises:
• (a) a corresponding plurality of input nodes, each respective input node in the corresponding plurality of input nodes for a different transcript abundance in the plurality of transcript abundance abundances, and
• (b) a representation of the corresponding gene set in the form of (i) a corresponding plurality of hidden nodes, each hidden node representing a gene in the corresponding gene set, and (ii) a corresponding plurality of edges, wherein each edge in the corresponding plurality of edges interconnects an input node in the plurality of input nodes to a hidden node in the corresponding plurality of hidden nodes with a corresponding edge weight;
• responsive to the inputting, obtaining a plurality of predictions, each prediction in the plurality of predictions from a neural network in the plurality of neural networks;
• and
• responsive to inputting the plurality of predictions into an ensemble model obtaining, as output form the ensemble model a prediction of whether the subject has the characteristic.

A method for determining one or more transcription factors that regulate a first gene in a cell type, the method comprising:

• A) obtaining a single nucleus multi-omics dataset, in electronic form, comprising:
• (i) a respective ATAC fragment count for each ATAC peak in a corresponding plurality of ATAC peaks, for each respective cell in a plurality of cells, and
• (ii) a respective discrete attribute value for each gene transcript in a corresponding plurality of gene transcripts, for each respective cell in the plurality of cells, wherein the plurality of cells is from a biological sample from a subject;
• B) obtaining a plurality of transcription factor binding sites, wherein each respective transcription factor binding site in the plurality of transcription factor binding sites is associated with (i) a gene in a plurality of genes and (ii) a transcription factor in a plurality of transcription factors;
• C) for each respective cell represented in the plurality of cells, for each respective transcription factor binding site in the plurality of transcription factor binding sites, using the respective ATAC fragment count for each corresponding ATAC peak from the respective cell in the single nucleus multi-omics dataset within a threshold distance of the respective transcription factor binding site to determine a respective binary openness assignment for the respective transcription factor binding site for the respective cell represented in the plurality of cells;
• D) for each respective cell represented in the plurality of cells, for each respective gene in the plurality of genes, wherein the plurality of genes includes the first gene, forming a respective regressor of form:
• z
• =
• f
• ⁡
• (
• y
• ij
• ·
• x
• i
• )
• wherein,
• z is the respective discrete attribute value of the respective gene for the respective cell in the single nucleus multi-omics dataset,
• xi is the respective discrete attribute value of the ith transcription factor associated with the respective gene for the respective cell in the single nucleus multi-omics dataset, and
• yij is the binary openness of the jth transcription factor binding site of the ith transcription factor in the respective cell,
• ƒ is a linear model, and
• i and j are positive integers,
• thereby forming a plurality of regressors;
• and
• E) regressing the plurality of regressors against the single nucleus multi-omics dataset, thereby identifying one or more transcription factors in the plurality of transcription factors that regulate the first gene.

A computer system for determining one or more transcription factors that regulate a first gene in a cell type, the computer system comprising:

• one or more processors;
• and
• memory addressable by the one or more processors, the memory storing at least one program for execution by the one or more processors, the at least one program comprising instructions for:
• A) obtaining a single nucleus multi-omics dataset, in electronic form, comprising:
• (i) a respective ATAC fragment count for each ATAC peak in a corresponding plurality of ATAC peaks, for each respective cell in a plurality of cells, and
• (ii) a respective discrete attribute value for each gene transcript in a corresponding plurality of gene transcripts, for each respective cell in the plurality of cells, wherein the plurality of cells is from a biological sample from a subject;
• B) obtaining a plurality of transcription factor binding sites, wherein each respective transcription factor binding site in the plurality of transcription factor binding sites is associated with (i) a gene in a plurality of genes and (ii) a transcription factor in a plurality of transcription factors;
• C) for each respective cell represented in the plurality of cells, for each respective transcription factor binding site in the plurality of transcription factor binding sites, using the respective ATAC fragment count for each corresponding ATAC peak from the respective cell in the single nucleus multi-omics dataset within a threshold distance of the respective transcription factor binding site to determine a respective binary openness assignment for the respective transcription factor binding site for the respective cell represented in the plurality of cells;
• D) for each respective cell represented in the plurality of cells, for each respective gene in the plurality of genes, wherein the plurality of genes includes the first gene, forming a respective regressor of form:
• z
• =
• f
• ⁡
• (
• y
• ij
• ·
• x
• i
• )
• wherein,
• z is the respective discrete attribute value of the respective gene for the respective cell in the single nucleus multi-omics dataset,
• xi is the respective discrete attribute value of the ith transcription factor associated with the respective gene for the respective cell in the single nucleus multi-omics dataset, and
• yij is the binary openness of the jth transcription factor binding site of the ith transcription factor in the respective cell,
• ƒ is a linear model, and
• i and j are positive integers,
• thereby forming a plurality of regressors;
• and
• E) regressing the plurality of regressors against the single nucleus multi-omics dataset, thereby identifying one or more transcription factors in the plurality of transcription factors that regulate the first gene.

A non-transitory computer readable storage medium, wherein the non-transitory computer readable storage medium stores instructions, which when executed by a computer system, cause the computer system to perform a method for determining one or more transcription factors that regulate a first gene in a cell type, the method comprising:

• A) obtaining a single nucleus multi-omics dataset, in electronic form, comprising:
• (i) a respective ATAC fragment count for each ATAC peak in a corresponding plurality of ATAC peaks, for each respective cell in a plurality of cells, and
• (ii) a respective discrete attribute value for each gene transcript in a corresponding plurality of gene transcripts, for each respective cell in the plurality of cells, wherein the plurality of cells is from a biological sample from a subject;
• B) obtaining a plurality of transcription factor binding sites, wherein each respective transcription factor binding site in the plurality of transcription factor binding sites is associated with (i) a gene in a plurality of genes and (ii) a transcription factor in a plurality of transcription factors;
• C) for each respective cell represented in the plurality of cells, for each respective transcription factor binding site in the plurality of transcription factor binding sites, using the respective ATAC fragment count for each corresponding ATAC peak from the respective cell in the single nucleus multi-omics dataset within a threshold distance of the respective transcription factor binding site to determine a respective binary openness assignment for the respective transcription factor binding site for the respective cell represented in the plurality of cells;
• D) for each respective cell represented in the plurality of cells, for each respective gene in the plurality of genes, wherein the plurality of genes includes the first gene, forming a respective regressor of form:
• z
• =
• f
• ⁡
• (
• y
• ij
• ·
• x
• i
• )
• wherein,
• z is the respective discrete attribute value of the respective gene for the respective cell in the single nucleus multi-omics dataset,
• xi is the respective discrete attribute value of the ith transcription factor associated with the respective gene for the respective cell in the single nucleus multi-omics dataset, and
• yij is the binary openness of the jth transcription factor binding site of the ith transcription factor in the respective cell,
• ƒ is a linear model, and
• i and j are positive integers,
• thereby forming a plurality of regressors;
• and
• E) regressing the plurality of regressors against the single nucleus multi-omics dataset, thereby identifying one or more transcription factors in the plurality of transcription factors that regulate the first gene.