Generating enhanced graphical user interfaces for presentation of anti-infective design spaces for selecting drug candidates
Inventors
Lee, Francis • STECKBECK, Jonathan D. • Holste, Hannes
Assignees
Publication Number
US-11967400-B2
Publication Date
2024-04-23
Expiration Date
2041-05-13
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
In one aspect, a method is disclosed for presenting, on a computing device, a graphical user interface (GUI) of a therapeutic tool. The method includes presenting, in a first screen of the GUI, a design space for a protein for an application, where the design space includes a set of sequences, where each sequence contains a respective set of activities pertaining to the application. The method also includes receiving, via a graphical element in the first screen, a selection of one or more query parameters of the design space, and presenting, in a second screen of the GUI, a solution space that includes a subset of the set of sequences, where each sequence contains the respective set of activities, where the subset of the set of sequences is selected based on the one or more query parameters.
Core Innovation
The invention provides a method and system for presenting, on a computing device, a graphical user interface (GUI) of a therapeutic tool specifically for exploring and selecting drug candidates based on enhanced anti-infective design spaces. This includes presenting a design space for proteins (such as peptides) where each protein sequence is linked to multiple activities relevant to a therapeutic application. The sequences in the design space are generated by a machine learning model that applies causal inference, simulating alternative scenarios to efficiently filter and select protein sequences from a larger superset.
The approach addresses problems in traditional drug discovery, which often suffers from limited design spaces, inefficiency, restriction by ‘known facts’, and the inability to handle extremely large and complex biological datasets. The patent describes how these conventional approaches overlook potentially valuable candidate drugs because they cannot scale with the complexity and multidimensionality required for finding new therapeutics.
By enabling interactive selection of query parameters via graphical elements in the GUI, users can filter the design space and view a solution space—comprising a subset of protein sequences with desirable activities—along with detailed, contextual information. The system further incorporates visualization techniques such as topographical maps and color-coded clusters, providing users with both high-level and granular data, including energy correlations, activities, pathways, and more, for each protein sequence in the solution space.
Claims Coverage
The patent claims cover a method, computer-readable medium, and apparatus for generating and presenting enhanced graphical user interfaces for visualizing and selecting protein sequences as candidate drug compounds using machine learning models with causal inference and interactive filtering.
Presenting a design space for proteins with activities generated by causal inference-based machine learning models
A design space containing a plurality of protein sequences for an application is presented on a GUI. Each sequence is associated with multiple activities. The sequences are generated by a machine learning model that uses causal inference to execute alternative scenarios, filtering a superset of protein sequences to construct the design space.
Interactive filtering using query parameters via GUI elements to derive a solution space
A graphical element in the first screen allows selection of one or more query parameters (including biomedical and non-biomedical ontology terms, characteristics, or descriptors) for the design space. Based on these parameters, the system presents in a second screen a solution space comprising a subset of the protein sequences, each retaining their associated set of activities.
Visualization of solution spaces with color-coded clusters and data-rich secondary displays
The second screen of the GUI shows color-coded clusters representing subsets of the protein sequences. A detailed second portion presents data about these subsets, describing associated objects such as drug compounds, activities, interactions, genes, pathways, physical descriptors, folding or wave properties, or stabilities. Selection mechanisms further allow presentation of unmatched or previously ungenerated sequences.
Tracking and presenting untraversed or novel protein sequences
The solution space distinguishes and highlights protein sequences not previously generated by the machine learning model, thereby facilitating the identification of novel candidates. The associated data objects are specifically linked to these untraversed protein sequences.
Providing a candidate dashboard with comprehensive sequence-specific data
Upon selection of a sequence in the solution space, a candidate dashboard is presented, showing structural details, correlation heatmaps, experimental data, probabilistic scores from inference models, and external related data for the candidate protein sequence.
Enabling trial execution leveraging the solution space with machine learning model performance metrics
Users can select a trial to be performed using the solution space. Results from the artificial intelligence engine include the point traversed in the solution space and metrics such as memory usage, GPU temperature, power usage, and processor usage, enabling evaluation of machine learning model resource efficiency.
Integration with business intelligence screen for target product profiles
A business intelligence screen allows users to input a target product profile (pharmacology, clinical, manufacturing, or other data). The system returns a subset of protein sequences matched to the profile and presents these within the GUI for further exploration and selection.
The claims encapsulate an interactive, machine learning-driven system for visualizing, filtering, and selecting protein sequences as candidate drug compounds, offering novel design spaces, detailed activity mapping, and integration with user-guided filtering and trial evaluation.
Stated Advantages
The approach enables efficient and scalable exploration of large, multidimensional biological design spaces, uncovering drug candidates that conventional methods would miss due to complexity or limited scope.
Users can interactively refine search criteria and immediately visualize the effects in the solution space, resulting in improved explainability and condensed, actionable information.
Visualization tools, such as topographical maps and color-coded clusters, enhance understanding of sequence activities and facilitate informed sequence selection through a single GUI.
The system supports identification and tracking of novel, untraversed, or previously ungenerated protein sequences, leading to improved discovery of unique candidate drugs.
Performance metrics help optimize resource utilization by comparing machine learning model executions, enabling improved computational resource management.
Integrated business intelligence tools permit matching of candidate sequences to specific target product profiles, allowing for rapid adaptation to business or clinical requirements.
Documented Applications
Presentation and exploration of design spaces for anti-infective drug discovery, including peptide and protein therapeutics.
Treatment development for diseases such as prosthetic joint infections, urinary tract infections, intra-abdominal infections, otitis media, cardiac infections, respiratory infections, neurological infections, dental infections, digestive and intestinal infections, and soft tissue infections.
Applications in therapeutic development for animal/veterinary health, including diseases like bovine mastitis.
Industrial applications such as anti-biofouling and optimized control sequences for machinery.
Application to other disease areas, including eczema, inflammatory bowel disease, Crohn's disease, rheumatoid arthritis, asthma, auto-immune diseases, and oncology treatments.
Enhancement of AI for generating sequences of decisions in video game non-player characters (NPCs).
Optimization of integrated circuit and chip design through improved mask works generation and routing processes.
Interested in licensing this patent?