The Predictive Phenomics Initiative (PPI) is comprised of 11 projects covering three main areas: microbial phenotyping, the science of scale-up, and artificial intelligence and automation. Now in its fifth year of operation, some PPI projects are complete, others are publishing results, and some are just beginning.

Here is a look at the FY26 projects:

Year 3

• Leveraging Phenomics to Improve Bioprocess Robustness (PI: Jeff Czajka)
• Variation-Leveraged Phenomic Association Study (PI: Jason McDermott)

Year 2

• Phenotypic Control via PTM Editing with Intrabodies (PIs: Elise Van Fossen/Samantha Powell)
• AI-Automation and Metabolic Modeling to Predict Phenotypes (PIs: Bram Stone/Joonhoon Kim)
• Transfer Learning for Experimental Design & Discovery (PIs: Kelly Stratton/Moses Obiri)
• Workflow for Reference Free Identification of Metabolites (Tricorder) (PI: Tom Metz)
• Computational Prediction of Small Molecule Targets that Control Cellular Phenotype (PI: John Melchior)
• Optimizing Bacterial-Fungal Biosystems for Phosphorus Mobilization and Organic Acid Production​ (PI: Sneha Couvillion)
• Predictive Phenomics Data Infrastructure (PI: Grant Fujimoto)

Year 1

• Foundational Autonomy for Biosystems Design (PI: Rob Egbert)
• Graph Inference for Microbial dark Matter Exploration (GIMME) (PI: Kaizad Patel)

Leveraging Phenomics to Improve Bioprocess Robustness

PI: Jeff Czajka

Goal:

• Identify and mitigate fundamental causes of bioprocess failures during fluctuations​ by investigating productivity phenotypes under reactor conditions.

Impact:

• Increased microbial process robustness at bench-scale will improve scale-up success and enable domestic bio-manufacturing of fuels and chemicals.

Variation-Leveraged Phenomic Association Study

PI: Jason McDermott 

Goals:

• Develop software that allows users to explore data-driven associations.
• Benchmark approach for existing datasets.
• Make predictions for important proteins and/or metabolites for yeast​.
• Provide novel functional annotations for unknown proteins and metabolites.

Impact:

• Ability to identify patterns in multi-omic data and provide meaningful functional annotations for unknown metabolites, proteins, and other molecules.

Phenotypic Control via PTM Editing with Intrabodies

PIs: Samantha Powell, Elise Van Fossen

Goal:

• Establish a novel capability for targeting functional PTMs to control phenotypes through generating nanobodies (Nbs) specific for a post-translationally modified proteoform of a given protein and targeting the PTM function by introducing the intracellular nanobodies (intrabodies)​.

Impacts:

• Creation of a tool to control the phenotypic outcome of intracelular PTMs without needing to knockout or permanently mutate the target​.
• Establish the first PTM-Nb selection process​.
• Create the first in vitro Nb selection process​.

AI-Automation and Metabolic Modeling to Predict Phenotypes 

PIs: Bram Stone/Joonhoon Kim 

Goals:

• Enhance current genome-informed metabolic model capabilities with relevant ecological and environmental parameters. 

Impacts:

• Advance the bioeconomy.
• Scale-up genome models toward community phenotype prediction.

Transfer Learning for Experimental Design and Discovery

PIs: Kelly Stratton & Moses Obiri

Goals:​

• Develop models with similar, pre-existing data to identify essential phenotype-influencing traits​.​
• Use transfer learning to apply models to ongoing projects at the experimental design stage to reduce the experimental space to explore​.​

Impact:​

• Model-guided experimental design for more efficient discovery of how to optimize phenotypes​.​

​

Workflow for Reference-Free Identification of Metabolites

PI: Tom Metz

Goal:

• Develop an integrated workflow for reference-free identification of novel metabolites and metabolic pathways in engineered organisms.

Impact:

• More efficient target production.

Computational Prediction of Small Molecule Targets That Control Cellular Phenotype

PI: John Melchior

Goal:

• Develop an advanced AI/ML tool that accurately predicts small molecule protein targets.

Impacts:

• Revolutionize drug discovery: Transform drug discovery and personalized medicine by designing treatments tailored to an individual’s unique biology, ensuring maximum efficacy with minimal off-target effects.
• Rapid response to emerging viruses: Enable rapid identification of antiviral compounds that protect host cells from new pathogens, ensuring a swift response to viral outbreaks.
• Enhance bioproduction: Pinpoint effective molecules that can modulate pathways to bolster bioproduction in microbial systems at scale.
• Optimize disease research: Empower researchers with promising molecules to gain deeper insights into disease mechanisms.
• Identify chemicals of concern: Rapidly assess potential positive or negative impacts on cellular responses of chemicals to enhance safety and efficacy in product development.

Optimizing Bacterial-Fungal Biosystems for Phosphorus Mobilization and Organic Acid Production​

PI: Sneha Couvillion

Goal: 

• Develop optimized bacterial-fungal biosystems for phosphorus (P) mobilization and organic acid production​.

​Impact​:

• A predictive framework that links measurable microbial traits in monoculture to emergent functional outcomes in bacterial-fungal co-cultures… enabling the design and optimization of biosystems with targeted biogeochemical functions​.

Predictive Phenomics Data Infrastructure

PI: Grant Fujimoto

Goal: 

• Design and build a web-based data infrastructure using industry standard practices to track, organize, and share PPI data and metadata. ​

Impacts​:

• Facilitates handoff of data between projects. ​

• Provides a single source of truth to organize, track, and store all PPI data. This could be exceptionally valuable to stats/modeling projects and simplify the publication process. ​

• The infrastructure creates a place for PPI data to live and remain accessible beyond the life of the infrastructure. ​

• Iterative process that will naturally lead to connection of projects​.
   

Foundational Autonomy for Biosystems Design

PI: Rob Egbert

Goal:

• Establish an autonomous science capability that's focused on bioeconomy.

Impact:

• Establish PNNL autonomous science capability for the bioeconomy​.

Graph Inference for Microbial Dark Matter Exploration

PI: Kaizad Patel

Goals:

• Use generative AI to accelerate discovery for microbial gene function​.
• Develop and standardize best practices for integrating ontology and annotations into knowledge graphs for biology​.
• Build a gen-AI workflow for generating gene/protein embeddings useful for autonomous experimentation objectives, specifically bioproduction in Pseudomonas. ​

Impact:

• Build a gen-AI workflow for generating gene/protein embeddings useful for autonomous experimentation objectives, specifically bioproduction in Pseudomonas. ​