Microbial Phenotyping With High-Content Imaging and AI Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for developing and optimizing microbial strains for biomanufacturing are limited by extensive biological knowledge requirements, costly experimental testing, and the inability to assess fitness and productivity in complex environments, often hiding effects of strain modifications in combinations.

Innovation Solution

A method using high-content imaging and computer-based models, such as deep learning, to predict microbial cell phenotypes from physical characteristics, enabling quick and non-invasive assessment of microbial fitness and productivity, and engineering cells with desired traits.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional methods (genotyping, measuring titers, measuring growth/survival) are used to assess microbial phenotypes, then measurement precision is improved, but loss of time and loss of substance increase

Engineering Contradiction:
Improvephenotype assessment accuracyVSAvoidtime for phenotyping
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent uses high-content imaging to create visual copies of microbial cells and their phenotypic characteristics. By capturing images of cells with known phenotypes and training machine learning models on these images, the system can predict phenotypes of new cells without performing time-consuming traditional assays. This copying approach allows rapid phenotyping while maintaining accuracy.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces traditional mechanical and chemical assay methods with computational modeling and machine learning algorithms. Instead of physically measuring titers or conducting growth assays, the system uses image processing and neural networks to predict phenotypes, significantly reducing the time and resources required while maintaining measurement precision.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If traditional methods are used to assess microbial phenotypes, then measurement precision is improved, but loss of substance increases

Engineering Contradiction:
Improvephenotype assessment accuracyVSAvoidresource consumption
Core Design Contradiction:
Measurement precisionVSLoss of substance

Solution Approach 1:

The patent creates digital copies of phenotypic information through high-content imaging. By capturing and storing visual data of microbial cells and their characteristics, the system eliminates the need for repeated physical assays and consumable reagents, thereby reducing substance loss while maintaining accurate phenotype assessment.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent substitutes physical and chemical measurement methods with computational analysis. By using machine learning models to process and interpret images, the system eliminates the need for expensive reagents, consumables, and labor-intensive procedures, significantly reducing resource consumption while preserving measurement precision.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Ease of operation

If reductionist approaches (testing one strain modification at a time) are used, then ease of operation is improved, but loss of information increases

Engineering Contradiction:
Improvesimplicity of testingVSAvoideffect of strain modifications in combinations
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The patent segments the complex task of phenotype assessment into multiple independent image features and characteristics. By capturing and analyzing individual visual attributes (cell morphology, staining patterns, spatial distribution), the machine learning model can integrate these segments to predict overall phenotypes, including interaction effects between multiple strain modifications without requiring systematic combination testing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent replaces systematic combination testing with computational pattern recognition. Machine learning algorithms automatically detect and interpret complex interactions between multiple strain modifications by analyzing visual patterns in high-content images, eliminating the need for exhaustive experimental combinations while preserving complete information about interaction effects.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

4Productivity

If high-content imaging with machine learning is used to predict phenotypes, then productivity is improved, but device complexity increases

Engineering Contradiction:
Improvephenotyping speedVSAvoidsystem complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent develops a universal phenotyping system that can predict multiple phenotype types (growth rate, titer, stress response, gene expression) using a single high-content imaging platform and machine learning framework. This multi-functional approach consolidates what would otherwise require multiple separate assessment systems into one integrated platform, managing complexity while maximizing productivity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent creates a digital twin system that captures phenotypic information through imaging and stores it as data models. By copying physical phenotypes into digital representations, the system enables rapid prediction and simulation without requiring complex physical measurement apparatus for each parameter, thereby managing device complexity while improving productivity.

Inventive Principle:
Principle #26Copying

Data Source

PatentUS20260055358A1Phenotypic and biological assessment of microbes
Publication Date: 2026.02.26 INSCRIPTA INC

AI summary

The present disclosure provides technologies for predicting a phenotype of a microbial cell using machine learning models trained using high-content imaging data (HCI). Also provided are methods of engineering a microbial cell to possess a phenotype of interest. Example phenotypes include the production of a target compound or biomolecule of interest. The provided technologies are useful for the efficient biomanufacturing of target compounds.