International Business Machines Corporation

Globalfoundries Inc.

Laxmi Priya Parida

Filippo Utro

David C Haws

Niina S Haiminen

Dan He

Kahn Rhrissorrakrai

Chaya Levovitz

Irina Rish

Alberto Apostolico

Matteo Comin

Gad M Landau

Oren Weimann

Aldo Guzman Saenz

Ajay K Royyuru

Gustavo Alejandro Stolovitzky

Isidore Rigoutsos

Jingwei Bai

Ruhong Zhou

Daniel E Platt

Aris Floratos

Asif Javed

Enrico Siragusa

Myson Burch

Aritra Bose

A computer-implemented method includes receiving at a processor an input set of multiway data. The method determines a set of persistent homology barcodes based on the multiway data and identifies at least a first significant persistent homology barcode in the set of persistent homology barcodes. A representative cycle of the first significant persistent homology is returned and an orthonormal basis for a boundary of the multiway data is computed. The method obtains a harmonic representative by computing a projection of the representative cycle to an orthogonal complement and generates a set of feature vectors using the harmonic representative. Each feature vector has a magnitude corresponding to an impact that feature has on a given condition. The method then ranks the feature vectors by feature vector magnitude.

Ranking cross-disorder features in multiway data interactions

A computer-implemented method includes to determine a cell, tissue or a lesion representation in cell-free DNA comprises inputting, to a processor, cell-free DNA (cfDNA) genomic profiles from one or more fluid biopsy samples from a patient and one or more genomic profiles from one or more cells, tissues or lesions from the patient; constructing, by the processor, a plurality of synthetic fluid hypotheses (SFs); comparing, by the processor, each of the plurality of SFs to the cfDNA genomic profiles to determine goodness of fit, of each of the plurality of SFs; selecting, by the processor, a subset of the plurality of SFs, wherein each SF of the subset of SFs has a minimum distance in goodness of fit compared to the cfDNA genomic profile; and outputting, by the processor, based on the subset of SFs, a cell, tissue or a lesion representation in the cfDNA of the patient.

Determining cell, tissue, or lesion representations in cell-free DNA

Methods and systems for training a machine learning model are described. A processor can transform single cell data in a first space into projection data in a second space having a dimensionality lower than or equal to the first space. The processor can produce a cover having a plurality of sets of the projection data. The processor can determine a plurality of transition paths among the plurality of sets. A transition path can represent a transition from one cell state to another cell state. The processor can translate the transition paths from the second dimensional space to the first dimensional space. The processor can extract features from the transition paths in the first dimensional space. The processor can generate training data using the features, and use the training data to train a machine learning model for classifying cell state transitions.

Cell state transition features from single cell data

A computer-implemented method is disclosed which includes receiving biological sample information from one or more subjects at a first time period. The method further includes receiving biological sample information from the one or more subjects at a second time period. The method further includes comparing the biological sample information at the second time period with the biological sample information at the first time period. The method further includes generating a precedence graph based on results of the comparison. The method further includes determining one or more actions based on the precedence graph.

Methods and systems for determining drug resistance using a precedence graph

A method includes generating, by a processor system, a graph. The graph is based at least in part on a plurality of instances in which operational taxonomic units are identified as being represented within an environment. The method can also include determining, using the processor system, that at least one instance of the plurality of instances corresponds to a false-positive identification of an operational taxonomic unit. The determining is based on the properties of the graph. The method can also include reporting the determination.

Analyzing metagenomics data

Various embodiments perform stable gene analysis of transcriptome sequencing data. In one embodiment, a plurality of datasets each including transcriptome sequencing data are received by a processor. Each of the plurality of datasets includes a plurality of genes and a respective ranking value for each of the plurality of genes. A plurality of rank normalized input datasets is generated based on assigning, for each of the plurality of datasets, a rank to each of the plurality of genes. One or more longest increasing subsequence (LIS) of ranks are identified between each pair of the plurality of rank normalized input datasets. A set of stable genes from the plurality of genes is identified based on each of the one or more LIS of ranks across the plurality of rank normalized input datasets.

Stable genes in comparative transcriptomics

Various embodiments select markers for modeling epistasis effects. In one embodiment, a processor receives a set of genetic markers and a phenotype. A relevance score is determined with respect to the phenotype for each of the set of genetic markers. A threshold is set based on the relevance score of a genetic marker with a highest relevancy score. A relevance score is determined for at least one genetic marker in the set of genetic markers for at least one interaction between the at least one genetic marker and at least one other genetic marker in the set of genetic markers. The at least one interaction is added to a top-k feature set based on the relevance score of the at least one interaction satisfying the threshold.

Feature selection for efficient epistasis modeling for phenotype prediction

A computer-implemented method includes generating, by a processor, a set of training data for each phenotype in a database including a set of subjects. The set of training data is generated by dividing genomic information of N subjects selected with or without repetition into windows, computing a distribution of genomic events in the windows for each of N subjects, and extracting, for each window, a tensor that represents the distribution of genomic events for each of N subjects. A set of test data is generated for each phenotype in the database, a distribution of genomic events in windows for each phenotype is computed, and a tensor is extracted for each window that represents a distribution of genomic events for each phenotype. The method includes classifying each phenotype of the test data with a classifier, and assigning a phenotype to a patient.

Single sample genetic classification via tensor motifs

A computer-implemented method incudes calculating, by a processor, based on sequence data for a tumor from a subject at a plurality of time points, a mutation frequency for each of a plurality of SSVs at each of the time points to provide a plurality of time-resolved mutation frequencies (between 0 and 1) for each of the plurality of SSVs, the sequence data including a plurality of simple somatic variations (SSVs) at each of the time points; binning, by the processor, the plurality of time-resolved mutation frequencies for each SSV at each of the time points to provide a matrix of SSVs and time points; converting, by the processor, the matrix cells to pseudo-clones; and constructing, by the processor, a time-series tumor evolution tree from the pseudo-clones, wherein each time point in the time-series evolution tree represents an event in the subject's cancer treatment.

Time-series phylogenetic tumor evolution trees

Embodiments of the invention include systems and methods for protecting study participant data for aggregate analysis. Aspects include sending a broker encryption key to a plurality of subjects. Aspects also include receiving double-encrypted subject data from the plurality of subjects. Aspects also include decrypting the double-encrypted subject data with a broker decryption key to generate single-encrypted subject data for the plurality of subjects. Aspects also include aggregating the single-encrypted subject data for the plurality of subjects to generate an aggregated single-homomorphically encrypted data set. Aspects also include including a plurality of random factors in the aggregated single-encrypted data set. Aspects also include sending the aggregated single-homomorphically encrypted data set to a researcher.

Protecting study participant data for aggregate analysis

A computer-implemented method includes determining, by a processor, from a time-series evolution tree comprising one or more clones at each of the plurality of time points, that the one or more clones are sensitive clones or resistant clones, wherein the time-series evolution tree is based on sequence data for a tumor from a subject at a plurality of time points, wherein each time point in the time-series evolution tree represents an event in the subject's cancer treatment, and wherein a clone is a collection of gene alterations; based at least in part on determining that the one or more clones that are the sensitive or resistant clones, determining, by the processor, a geneset composition of the one or more clones that are the sensitive or resistant clones; and based at least in part on determining the geneset composition, determining by the processor, a further treatment for the subject.

Functional analysis of time-series phylogenetic tumor evolution tree

Embodiments include methods, systems, and computer program products for analyzing genomic data. Aspects include receiving genomic data for an organism, sample phenotypes, and a plurality of gene sets. Aspects include, for each of the gene sets, determining a set of genes G corresponding to genes in the gene set and a set of genes G' corresponding to genes outside the gene set for the phenotypes R and R&#x2032;. Aspects also include determining a set of mutated genes M and a set of non-mutated genes M&#x2032; for R and R&#x2032; and a mutation enrichment score. Aspects also include determining a set of differentiated genes D a set of non-differentiated genes D&#x2032; for R and R&#x2032;. Aspects also include identifying an enriched gene set Gbased at least in part upon the mutation enrichment score and the differentiation enrichment score.

Differential gene set enrichment analysis in genome-wide mutational data

A computer-implemented method, computer program product, and computer processing system are provided for metagenomic pattern classification. The method includes pre-processing, by a processor, a taxonomy tree associated with a genome database to extract taxonomy related information therefrom. The genome database includes a plurality of genome sequences. The method further includes building, by the processor, a suffix tree on the genome database. The method also includes annotating, by the processor, nodes in the suffix tree, using a plurality of right maximal patterns derived from the extracted taxonomy related information as annotations, such that each of the plurality of right maximal patterns in the suffix tree points to a respective one of a plurality of nodes in the taxonomy tree and such that a leaf node in the taxonomy tree represents a respective sample organism. The annotations are configured to function as classifications for the plurality of genome sequences.

Growing community of inventors

Mohegan Lake, NY, United States of America

Laxmi Priya Parida