Searching biology for the next generation of antimicrobials.
Zemnas Biosciences is building Lysora, an AI-guided antimicrobial discovery engine designed to identify and prioritize novel antimicrobial peptide candidates against drug-resistant pathogens.
FIELD 1.2M SEQ
MODE SEQUENCE SPACE
Antimicrobial resistance needs a larger search space.
Resistance
A rising biological threat outpacing conventional discovery.
Discovery
Traditional pipelines are slow and increasingly narrow.
Sequence space
Vast, structured, and largely unexplored by experiment.
Estimated deaths associated with antimicrobial resistance each year.
Roughly one every six seconds, while the pipeline of genuinely new antimicrobial classes has stayed thin for decades. The search space itself has to grow.
Reference estimate · GRAM 2022, The LancetNature has evolved enormous molecular diversity. We believe computational biology can help uncover antimicrobial candidates hidden within it.
Nature's oldest antimicrobials, searched with new tools.
Antimicrobial peptides are part of innate immunity across nearly all life: short, potent, and structurally diverse. Exactly the kind of needles a computational engine is built to find.
Compact by design
Short chains are easier to reason about computationally, and often easier to optimize as therapeutics.
A familiar alphabet
The same amino acids biology already uses, rearranged into sequences it has never been shown.
An ocean of candidates
Even a modest peptide length implies a search space no laboratory could screen by hand. That is precisely where computation earns its place.
Harder to out-evolve
Many AMPs attack the physical structure of the membrane itself, a target pathogens struggle to mutate away.
Lysora combines biological sequence mining, machine learning and computational validation to transform large biological search spaces into prioritized antimicrobial peptide candidates.
Discover
Mine biological sequence space for candidate peptides.
Evaluate
Estimate antimicrobial potential and target-specific potency.
De-risk
Assess safety, uncertainty and developability.
Prioritize
Advance the strongest candidates toward experimental validation.
Not one model. A discovery system.
Sequence discovery
Activity evidence
Target-specific potency
Safety
Mechanistic evidence
Uncertainty
Candidate prioritization
The prediction is only the beginning.
Computational Discovery
Candidate Prioritization
Experimental Validation
Lead Development
Candidates are designed to move beyond computational prediction into experimental testing, iterative learning and therapeutic development.
Initial development is focused on target-specific antimicrobial activity and potency against clinically relevant resistant pathogens, including MRSA / S. aureus.
Discovery designed to create therapeutics.
Discovery partnerships
Collaborative target-specific peptide discovery with research and biotechnology partners.
Co-development
Shared programs that move prioritized candidates through validation and optimization.
Candidate licensing
Proprietary peptide assets available for therapeutic development and licensing.
Zemnas is building proprietary antimicrobial peptide programs while exploring collaborations with researchers, biotechnology companies and therapeutic development partners.
Every candidate that leaves the engine carries both ideas: something old, borrowed from biology, and something new, found by computation.
The next antimicrobial may already exist somewhere in biology.
Lysora searches biological sequence space for antimicrobial potential.
A computational discovery engine integrating bioinformatics, machine learning, target-specific activity modelling, safety assessment and uncertainty-aware candidate prioritization.
A candidate takes shape.
SEQUENCE · KLLKWLLKDesigned to engage the membrane.
From raw sequence to a ranked candidate.
Discover
Mine proteins, genomes and biological sequence collections for potential antimicrobial peptides.
Predict
Evaluate antimicrobial potential using benchmarked protein-language and physicochemical models.
Target
Estimate target-specific potency against selected pathogens.
De-risk
Evaluate hemolysis, developability, domain familiarity and predictive uncertainty.
Prioritize
Integrate computational evidence to select candidates for experimental validation.
Six questions every candidate must answer.
Does it look antimicrobial?
Sequence-level activity evidence from benchmarked protein-language models.
Against what?
Target-specific potency estimates against priority pathogens, not generic averages.
Is it safe?
Hemolytic risk, cytotoxicity signals and host-cell selectivity, assessed in silico.
How might it work?
Mechanistic plausibility through membrane-interaction modelling and simulation.
How sure are we?
Explicit uncertainty quantification: low-confidence predictions are down-ranked, not hidden.
Can it be developed?
Stability, synthesis and developability heuristics before anything reaches a wet lab.
Computational evidence narrows the search. Experimental biology determines what advances.
What Lysora is not.
Not a black box
Every candidate carries its evidence trail: activity, potency, safety and uncertainty, each inspectable on its own.
Not a replacement for the lab
Predictions are ranked hypotheses. Assays are the arbiter, and the pipeline is built to feed them, not bypass them.
Not a single model
Multiple evidence layers, deliberately independent, so no single failure mode can quietly dominate a decision.
Multiple evidence layers, one decision.
From sequence to therapeutic candidate.
Lysora is being developed as an iterative discovery pipeline connecting computational prediction with experimental validation.
Biological sequence mining
Discover potential peptide regions across biological sequence space.
Computational prioritization
Rank candidate peptides using activity, physicochemical and uncertainty-aware models.
Target-specific potency
Model antimicrobial potency against priority pathogens including S. aureus and MRSA.
Safety & selectivity
Evaluate hemolytic risk and host-cell selectivity.
Mechanistic validation
Investigate membrane interaction and structural evidence using molecular modelling and simulation.
Experimental validation
Validate prioritized peptides through antimicrobial and safety assays.
Lead optimization
Use experimental evidence to guide peptide optimization and iterative discovery.
What "validated" actually means.
No percentages, no theatrics. A candidate earns its place through three concrete experiments, and every result, positive or negative, becomes training evidence for the next cycle.
Antimicrobial activity
Minimum inhibitory concentration against priority strains, including MRSA.
Hemolysis
Red-blood-cell safety window, the first selectivity gate any peptide must pass.
Cytotoxicity
Host-cell viability across a concentration range, not a single dose.
Every experiment teaches the next search.
Experimental successes and failures become new evidence for the next discovery cycle: Design, Build, Test, Learn, turning validation into a compounding advantage.
Biology has already evolved extraordinary molecular solutions. We want to find them.
Zemnas Biosciences is an early-stage biotechnology company combining artificial intelligence, computational biology and peptide science to discover new approaches to antimicrobial resistance.
Drug resistance continues to reduce the effectiveness of existing antimicrobial therapies.
Biological sequence space contains enormous unexplored molecular diversity.
Use computation to search more intelligently, then let experiments decide what advances.
The motifs worth mining.
Across unrelated organisms, certain peptide patterns recur again and again, preserved because they work. Lysora learns to read those conserved columns and treat them as a map of where to look next.
Built step by step, in the open where it matters.
The question
Why does antimicrobial discovery keep missing the largest molecular library in existence, the one biology already wrote? Zemnas is founded around that question.
Lysora v0.1
First sequence-mining engine assembled: discovery models running over biological sequence collections, with an evidence-first ranking philosophy from day one.
Evidence layers
Activity, safety and uncertainty layers integrated into a single prioritization system, so predictions stop being scores and start being dossiers.
Target-specific potency
Modelling sharpened toward priority pathogens, with S. aureus and MRSA as the first program focus.
Experimental validation
Prioritized candidates move into antimicrobial and safety assays, in collaboration with wet-lab partners. This is the milestone we are building toward right now.
Three principles that hold the work together.
Scientific honesty
Predictions are hypotheses, not proof. We report evidence, not certainty.
Computational scale
Search biological spaces impossible to explore manually, then surface what matters.
Experimental grounding
Every candidate ultimately has to survive biological validation.
Building at the intersection of biology and computation.
Working across computational biology, AI-guided drug discovery and antimicrobial peptide discovery to build Zemnas Biosciences and the Lysora platform.
Research & Scientific Network
Zemnas is building relationships with antimicrobial researchers, microbiology laboratories and peptide scientists. Future collaborators and advisors will appear here with permission.
Let's validate what computation discovers.
Zemnas is looking to collaborate with antimicrobial researchers, microbiology laboratories, peptide scientists and therapeutic development partners.
Experimental validation
- MIC / antimicrobial activity
- Hemolysis
- Cytotoxicity
- Mechanistic assays
Scientific collaboration
- AMR biology
- Peptide science
- Membrane biology
- Target selection
Discovery partnerships
- Target-specific peptide discovery
- Co-development
- Candidate licensing
Tell us what you can test.
hello@zemnasbio.comWe read every message. Whether you run antimicrobial assays, study membrane biology, or develop peptide therapeutics, we would like to hear from you.