# US 2026/0066047 A1 · Models and methods for predicting cell-surface protein expression

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0066047 A1
- Document ID: US20260066047A1
- Status: Published application
- Kind: A1
- Application number: 19/317,440
- Filed: 2025-09-03
- Published: 2026-03-05
- Earliest claimed: 2024-09-04
- Assignees: Wisconsin Alumni Research Foundation
- Applicants: Wisconsin Alumni Research Foundation
- Inventors: Shuang Zhao
- Web page: https://patentgrep.com/patent/US20260066047A1

## CPC

- G16B 25/10: Gene or protein expression profiling; Expression-ratio estimation or normalisation
- G16B 5/00: ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
- G16B 45/00: ICT specially adapted for bioinformatics-related data visualisation, e.g. displaying of maps or networks
- G16B 50/30: Data warehousing; Computing architectures
- G16H 10/40: for data related to laboratory analysis, e.g. patient specimen analysis
- G16H 10/60: for patient-specific data, e.g. for electronic patient records
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA

## Abstract

Models and methods for predicting cell-surface protein expression, such as expression of cell-surface targets on cancer cells.

## Claims

### Claim 1

A method of generating a prediction model of cell-surface protein expression in cancer, the method comprising:
determining a gene expression profile for each training cell sample in a set of training cell samples, wherein each gene expression profile comprises a set of gene-expression values for a set of gene-expression-profile genes, wherein the set of training cell samples comprises a set of training cancer cell samples;
determining a set of training genes, wherein the set of training genes comprises a common set of gene-expression-profile genes represented in each of the gene expression profiles, wherein the set of training genes comprises a set of cell-surface protein genes;
ranking the gene-expression values for all of the training genes in each expression profile to thereby obtain a training ranking for each gene expression profile, wherein each training ranking comprises a rank for the gene-expression value for each training gene within the gene expression profile;
identifying, for each training gene, the training cell samples having a same rank.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
generating a matrix data structure comprising a plurality of rows as the training cell samples and columns as the cancer cell samples; and
storing the matrix data structure in computer memory.

### Claim 3 (depends on 2)

The method of claim 2, wherein the matrix data structure is stored in contiguous memory in RAM.

### Claim 4 (depends on 2)

The method of claim 2, further comprising:
communicating the matrix data structure from the computer memory to a networked database.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
prior to determining the gene expression profile for each training cell sample, pre-processing the set of training cell samples, wherein pre-processing comprises normalizing for gene length and sequencing depth.

### Claim 6 (depends on 5)

The method of claim 5, wherein the normalizing is based on at least one of Transcripts Per Kilobase Million (TPM), Fragments Per Kilobase Million (FPKM), or Reads Per Kilobase Million (RPKM).

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
prior to ranking the gene-expression values for all of the training genes in each expression profile, normalizing for gene length and sequencing depth.

### Claim 8 (depends on 7)

The method of claim 7, wherein the set of training cell samples is missing at least one piece of data, and wherein the normalizing accounts for the missing at least one piece of data.

### Claim 9 (depends on 1)

The method of claim 1, wherein the ranking comprises ordinally ranking the gene-expression values for all of the training genes in each gene expression profile to thereby obtain an ordinal training ranking for each gene expression profile, wherein each ordinal training ranking comprises an ordinal rank for the gene-expression value for each training gene within the gene expression profile.

### Claim 10 (depends on 9)

The method of claim 9, wherein the ranking further comprises:
determining a number of genes in the set of training genes; and
transforming each ordinal ranking into a quantile ranking by dividing the ordinal rank for the gene-expression value for each gene within the gene expression profile by the number of genes in the set of training genes.

### Claim 11 (depends on 1)

The method of claim 1, wherein the set of training cell samples further comprises training non-cancer cell samples.

### Claim 12 (depends on 1)

The method of claim 1, wherein the training cell samples comprise single cell samples.

### Claim 13 (depends on 12)

The method of claim 12, wherein the single cell samples comprise single cancer cells and single non-cancer cells.

### Claim 14

A method of using a prediction model to determine a predicted cell-surface target on a patient cancer cell sample, the method comprising:
ranking gene-expression values of all of a plurality of training genes in the patient cancer cell sample to obtain a patient ranking;
selecting one or more training genes having a threshold rank in the patient ranking to thereby obtain one or more patient genes;
comparing the rank(s) of the one or more patient genes in the patient cancer cell sample with the rank(s) of the one or more patient genes in the training cell samples to determine a number, type, and/or proportion of training cancer cell samples having the same rank(s) for the one or more patient genes; and
optionally, empirically testing cell-surface expression of the one or more patient genes in a test cancer cell sample from the patient.

### Claim 15 (depends on 14)

The method of claim 14, further comprising:
prior to ranking the gene-expression values for all of the training genes in each expression profile, normalizing for gene length and sequencing depth.

### Claim 16 (depends on 15)

The method of claim 15, wherein the set of training cell samples is missing at least one piece of data, and wherein the normalizing accounts for the missing at least one piece of data.

### Claim 17 (depends on 14)

The method of claim 14, further comprising:
presenting a graphical user interface to a user of the compared rank(s) of the one or more patient genes in the patient cancer cell sample with the rank(s) of the one or more patient genes in the training cell samples as a plurality of top cell surface targets organized by percentile.

### Claim 18 (depends on 17)

The method of claim 17, further comprising:
generating a pseudo-dynamic graphical user interface that simulates a dynamic user experience by:
pre-generating an image of all possible integer percentiles for a given training cancer cell as a plurality of images; and
populating the graphical user interface with a given image based on user input of an inputted training gene and an inputted percentile.

### Claim 19 (depends on 18)

The method of claim 18, wherein the graphical user interface is populated with a given image without using an index data structure.

### Claim 20 (depends on 18)

The method of claim 18, wherein each of the plurality of images comprise a filename including training gene and percentile.
