Patents by Inventor Collin A. Melton
Collin A. Melton has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20240136018Abstract: Methods and systems are disclosed for component deconvolution by a mixture model based on methylation information. A mixture model may be trained agnostic of labels or known component contributions. A system generates a methylation signature for each of a plurality of training samples. The methylation signature may be based on a count or a percentage of a methylation variant(s) expressed in the methylation sequence reads of a training sample at each genomic region of a plurality of genomic regions. The system may train the mixture model using maximum likelihood estimation to deconvolve the component contributions. The mixture model may comprise component submodels and a deconvolution submodel. The component submodels predict a component likelihood based on the methylation signature. The deconvolution submodel predicts the component contributions based on the component likelihoods.Type: ApplicationFiled: October 17, 2023Publication date: April 25, 2024Inventors: Aaron Stern, Joerg Bredno, Joseph Marcus, Oliver Claude Venn, Collin Melton
-
Publication number: 20240062849Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.Type: ApplicationFiled: August 31, 2023Publication date: February 22, 2024Applicant: GRAIL, LLCInventors: Virgil NICULA, Anton VALOUEV, Darya FILIPPOVA, Matthew H. LARSON, M. Cyrus MAHER, Monica Portela dos Santos Pimentel, Robert Abe Paine CALEF, Collin MELTON
-
Patent number: 11783915Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.Type: GrantFiled: September 29, 2022Date of Patent: October 10, 2023Assignee: GRAIL, LLCInventors: Virgil Nicula, Anton Valouev, Darya Filippova, Matthew H. Larson, M. Cyrus Maher, Monica Portela dos Santos Pimentel, Robert Abe Paine Calef, Collin Melton
-
Publication number: 20230272486Abstract: A computer-implemented method for generating a tumor fraction estimate from a DNA sample of a subject is disclosed. The method may include receiving a dataset of methylation sequence reads from the sample of the subject. The method may also include dividing the dataset into a plurality of variants. The method may further include determining methylation states of the plurality of variants. The method may further include filtering the plurality of variants based on a bank of reference sequence reads to generate a filtered subset of variants. The bank may include reads generated from non-cancer samples and biopsy samples of a plurality of tissues of reference individuals. The counts of the methylation states of variants in the filtered subset are determined and input to a model that is trained based on recurrence rates of the variants in the reference sequence reads. The tumor fraction estimate may be generated by the model.Type: ApplicationFiled: February 15, 2023Publication date: August 31, 2023Inventors: Collin Melton, Archana S. Shenoy, Joerg Bredno, Oliver Claude Venn, Konstantin Davydov, Matthew H. Larson
-
Publication number: 20230045925Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.Type: ApplicationFiled: September 29, 2022Publication date: February 16, 2023Applicant: GRAIL, LLCInventors: Virgil Nicula, Anton Valouev, Darya Filippova, Matthew H. Larson, M. Cyrus Maher, Monica Portela dos Santos Pimentel, Robert Abe Paine Calef, Collin Melton
-
Patent number: 11482303Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.Type: GrantFiled: May 31, 2019Date of Patent: October 25, 2022Assignee: GRAIL, LLCInventors: Virgil Nicula, Anton Valouev, Darya Filippova, Matthew H. Larson, M. Cyrus Maher, Monica Portela dos Santos Pimentel, Robert Abe Paine Calef, Collin Melton
-
Publication number: 20210292845Abstract: Systems and methods of identifying methylation patterns discriminating or indicating a cancer condition are provided. First and second datasets are obtained. Each dataset comprises a plurality of fragment methylation patterns determined by methylation sequencing of nucleic acids obtained from a first or second set of subjects and comprising a methylation state of each CpG site in a corresponding plurality of CpG sites. Each plurality of subjects has a respective first or second state of the cancer condition. First and second interval maps are generated for each respective dataset, each comprising a plurality of nodes characterized by a start methylation site, an end methylation site, a representation of each different fragment methylation pattern and a count of fragments.Type: ApplicationFiled: February 26, 2021Publication date: September 23, 2021Inventors: Collin Melton, Earl Hubbell, Oliver Claude Venn
-
Publication number: 20210285042Abstract: An allelic position variant calling method using a prior genotype probability at the allelic position is provided. A strand specific base count set in forward and reverse directions for the allelic position is obtained, using strand orientation and identity of a respective base at the allelic position in each respective nucleic acid fragment sequence that maps to the allelic position, where bases at the allelic position whose identity can be affected by conversion of cytosine to uracil do not contribute to the strand specific base count set. Respective forward and reverse strand conditional probabilities are computed for each candidate genotype for the allelic position using the strand specific base count set and sequencing error estimate. Likelihoods are computed using a combination of these conditional probabilities and the prior genotype probability. From this, a determination is made as to whether the likelihoods support a variant call at the allelic position.Type: ApplicationFiled: February 25, 2021Publication date: September 16, 2021Inventors: Pranav Singh, Christopher Chang, Collin Melton, Oliver Claude Venn
-
Publication number: 20210134394Abstract: A system and method for determining a presence of cancer in a test sample from a test subject comprising a set of fragments of deoxyribonucleic acid (DNA) is described. Locations along a genome of the test subject that are predictively significant in cancer detection may be identified through probabilistic analyses based on a comparison of the count of non-cancer fragments expected to terminate at a location and a count of fragments observed to terminate at the location. Based on the comparison, a p-value for each location is determined and is compared to a p-value threshold to determine predictively significant genomic locations, and a classifier is trained based on these locations. The system inputs a test feature vector containing counts of endpoint fragments from a test sample to the classifier, which generates a cancer prediction describing a likelihood the test sample has cancer and/or is of a particular cancer type.Type: ApplicationFiled: October 9, 2020Publication date: May 6, 2021Inventors: Peter D. Freese, Collin Melton, Earl Hubbell
-
Publication number: 20210102262Abstract: Systems and methods for determining whether a subject has a disease condition in a set of disease conditions are provided. The method includes obtaining a test dataset that comprises a first plurality of bin values obtained for a first plurality of bins collectively representing a first portion of a reference genome, and a second plurality of bin values obtained for a second plurality of bins collectively representing a second portion of the reference genome. The first and second plurality of bin values are derived from a targeted sequencing of a plurality of nucleic acids that are enriched using a plurality of probes. A plurality of copy number values are determined from the first and second plurality of bin values. The copy number values are inputted into a trained classifier, thereby determining whether the subject has a disease condition.Type: ApplicationFiled: September 16, 2020Publication date: April 8, 2021Inventors: Anton Valouev, Jing Xiang, Collin Melton
-
Publication number: 20200340064Abstract: Systems and methods for cancer subject tumor fraction estimation comprise obtaining a first plurality of nucleic acid fragment sequences from the subject's liquid biological sample. The first plurality of sequences represent cell-free nucleic acids in the liquid sample. A second plurality of nucleic acid fragment sequences is obtained from the subject's tumor sample. The second plurality of sequences represent nucleic acid molecules in the tumor. Smoothed noise rates, each determined using nucleic acid fragment sequences from non-cancer samples mapping to a corresponding allele position in a plurality of allele positions, are obtained. Variant allele counts and coverages are determined for the allele positions using the first plurality of sequences. Solid variant allele fractions are determined for the plurality of allele positions using the second plurality of sequences.Type: ApplicationFiled: April 16, 2020Publication date: October 29, 2020Inventors: Samuel S. Gross, Joshua Newman, Pranav Parmjit Singh, Collin Melton, Oliver Claude Venn, Earl Hubbell
-
Publication number: 20200051662Abstract: An approach to designing a polynucleotide probe to hybridize selectively to a target polynucleotide sequence involves calculating the final concentration of the intended binding product between a candidate probe and the target sequence. The calculation takes into consideration the binding reaction between the candidate probe and the target fragment on the target sequence, as well as various other binding reactions, involving either the probe or the target fragment, that interfere with the intended binding reaction. In contrast to the conventional technology, which attempts to determine the entire structure of the target polynucleotide, this approach only needs to determine the binding dynamics that impact on the intended probe-target fragment binding. The approach does not require determination of the structure of the involved sequences.Type: ApplicationFiled: October 25, 2019Publication date: February 13, 2020Inventors: Brian M. Frezza, Bradley M. Bond, Collin A. Melton, Catherine L. Hofler, Daniel J. Kleinbaum
-
Publication number: 20200005899Abstract: Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.Type: ApplicationFiled: May 31, 2019Publication date: January 2, 2020Inventors: Virgil Nicula, Anton Valouev, Darya Filippova, Matthew H. Larson, M. Cyrus Maher, Monica Portela dos Santos Pimentel, Robert Abe Paine Calef, Collin Melton
-
Patent number: 10504612Abstract: An approach to designing a polynucleotide probe to hybridize selectively to a target polynucleotide sequence involves calculating the final concentration of the intended binding product between a candidate probe and the target sequence. The calculation takes into consideration the binding reaction between the candidate probe and the target fragment on the target sequence, as well as various other binding reactions, involving either the probe or the target fragment, that interfere with the intended binding reaction. In contrast to the conventional technology, which attempts to determine the entire structure of the target polynucleotide, this approach only needs to determine the binding dynamics that impact on the intended probe-target fragment binding. The approach does not require determination of the structure of the involved sequences.Type: GrantFiled: March 13, 2013Date of Patent: December 10, 2019Assignee: EMERALD THERAPEUTICS, INC.Inventors: Brian M. Frezza, Bradley M. Bond, Collin A. Melton, Catherine L. Hofler, Daniel J. Kleinbaum