Microsoft Technology Licensing, LLC

Jennifer Listgarten

David Earl Heckerman

Carl Myers Kadie

Nicolo Fusi

Nebojsa Jojic

Noah Aaron Zaitlen

Gregory Byer Darnell

In embodiments of latent space harmonization (LSH) for predictive modeling, different training data sets are obtained from different measurement methods, where input data among the training data sets is quantifiable in a common space but a mapping between output data among the training data sets is unknown. A LSH module receives the training data sets and maps a common supervised target variable of the output data to a shared latent space where the output data can be jointly yielded. Mappings from the shared latent space back to the output training data of each training data set are determined and used to generate a trained predictive model. The trained predictive model is useable to predict output data from new input data with improved predictive power from the training data obtained using various, otherwise incongruent, measurement techniques.

Latent space harmonization for predictive modeling

This disclosure presents a model for identifying correlations in genome-wide association studies (GWAS) with function-valued traits that provides increased power and computational efficiency by use of a Gaussian process regression with radial basis function (RBF) kernels to model the function-valued traits and specialized factorizations to achieve speed. A Gaussian Process is assigned to each partition for each allele of a given single nucleotide polymorphism (SNP) which yields flexible alternative models and handles a large number of data points in a way that is statistically and computationally efficient. This model provides techniques for handling missing and unaligned function values such as would occur when not all individuals are measured at the same time points. If the data is complete algebraic re-factorization by decomposition into Kronecker products reduces the time complexity of this model thereby increasing processing speed and reducing memory usage as compared to a naive implementation.

Computationally efficient correlation of genetic effects with function-valued traits

A system described herein includes a receiver component that receives an HLA data set, wherein the HLA data set comprises low resolution HLA data. An HLA refinement component comprises a statistical model that automatically refines the HLA data set to transform the low resolution HLA data to high resolution HLA data.

Refining HLA data

Epitope prediction models are described herein. By way of example, a system for predicting epitope information relating to a epitope can include a classification model (e.g., logistic regression model). The trained classification model can illustratively operatively execute one ore logistic functions on received protein data, and incorporate one or more of hidden binary variables and shift variables that when processed represent the identification (e.g., prediction) of one or more desired epitopes. The classification model can be configured to predict the epitope information by processing data including various features of an epitope, MHC, MHC supertype, and Boolean combinations thereof.

T-cell epitope prediction

The claimed subject matter provides systems and/or methods that determines a number of non-spurious arcs associated with a learned graphical model. The system can include devices and mechanisms that utilize learning algorithms and datasets to generate learned graphical models and graphical models associated with null permutations of the datasets, ascertaining the average number of arcs associated with the graphical models associated with null permutations of the datasets, enumerating the total number of arcs affiliated with the learned graphical model, and presenting a ratio of the average number of arcs to the total number of arcs, the ratio indicative of the number of non-spurious arcs associated the learned graphical model.

Growing community of inventors

Redmond, WA, United States of America

Jennifer Listgarten