Methods and systems for predicting membrane protein expression based on sequence-level information
AssigneeCalifornia Institute of Technology
InventorsShyam M. SALADI, William M. CLEMONS
Life of this patent
Estimating the term…
Abstract
Membrane protein expression can be predicted using statistical frameworks to provide an enhanced subset of sequences, out of an initial larger set of potentially expressing sequences, by using features derived from sequences and a model parameterized through a dataset of known expression levels. Also, membrane protein experimentation protocols can be designed using the statistical frameworks in concert known outcomes to identify which laboratory conditions are most likely to produce successful results.
Claims28
A method for membrane protein expression, comprising:
- selecting an initial set of sequences with known protein expression for a membrane protein;
- transforming the initial set of sequences to features including nucleotide features, protein features, or both;
- dividing the initial set of sequences into a training set and a test set;
- creating a model based on the training set, the model providing statistical analysis of expressing a membrane protein;
- validating the model with the test set;
- using the model on a distinct set of sequences;
- creating an experimentation set of sequences by selecting a subset of the input set of sequences, wherein the selection is based on the results of the using the model;
- performing experimental testing on the subset such that expression levels are determined for each member of the subset, creating a positive result set of sequences that have positive expression of the membrane protein as determined by the experimental testing;
- and
- using the positive result set to express the membrane protein.
A method for designing membrane protein expression experimental protocol, comprising:
- selecting a membrane protein of interest;
- identifying expressed similar proteins in a plurality of conditions;
- for each of the plurality of conditions:
- selecting an initial set of sequences with known protein expression for a membrane protein,
- transforming the initial set of sequences to features including nucleotide features, protein features, or both,
- dividing the initial set of sequences into a training set and a test set, creating a model based on the training set, the model providing statistical analysis of expressing the membrane protein, and
- validating the model with the test set;
- determining which of the plurality of conditions has a best predictive performance based on the validating for each of the plurality of conditions;
- and
- performing membrane protein expression experiments on the membrane protein of interest under those of the plurality of conditions that have the best predictive performance.
Dependencies are inferred from the claim text.