Aliquant.bio

Computational biotech, run by agents

The first fully agentic computational biotechnology company.

Aliquant Bio deploys autonomous AI agents that read the literature, pick targets, design molecules and run in-silico experiments — starting with antivirals for hand, foot & mouth disease in young children.

See the HFMD pipeline Read the Substack →
agent-orchestrator

        

      

    


    
    

      
0approved antivirals for HFMD in the US or EU

      
<5 yrsthe age group most affected

      
EV‑A71serotype behind most severe neurological cases

      
24/7agents iterating on design–test cycles

    


    
    

      
Platform

      
An autonomous design–make–test loop, in silico.

      
Specialised agents hand work to each other, with every decision logged and reviewable by a human.

      
    
        
  1. 
          01
          
    Read
    
          
    Literature agents mine papers, patents and structural databases to map what is known about each viral target.
    
        
    2. 
        
  2. 
          02
          
    Hypothesise
    
          
    Reasoning agents rank targets and binding pockets, and propose what to test next and why.
    
        
    3. 
        
  3. 
          03
          
    Design
    
          
    Generative and ML scoring models propose, filter and optimise candidate molecules.
    
        
    4. 
        
  4. 
          04
          
    Validate
    
          
    Docking, ADMET prediction and critique agents stress-test candidates before anything reaches a wet lab.
    
        
    5. 
      

    


    
    

      
Pipeline

      
Hand, foot & mouth disease.

      

        HFMD causes large seasonal outbreaks among young children, especially across the Asia‑Pacific region.
        Most cases are mild, but enterovirus A71 can cause severe neurological disease — and there is still no
        approved antiviral treatment. That is the gap we are going after first.
      

      

        

          Program
          Target
          
            Target IDHit discoveryLead opt.Preclinical
          
        

        

      

      
Programs shown are placeholders — edit the PIPELINE list in main.js.

    


    
    

      
Machine learning

      
The models behind the agents.

      

        Our agents are only as good as the models they call — so we build and fine‑tune our own,
        each aimed at a step that off‑the‑shelf tools don't handle well.
      


      

        

          

            
Aligene‑1

            Live · v1
          

          
DNA sequence model · post‑trained for manufacturability

          

            Aligene‑1 starts from a DNA foundation model and is post‑trained for
            manufacturability — it doesn't just propose sequences, it scores and optimises
            them for what can actually be made. The agents use it to keep designs synthesisable and
            expressible before anything reaches a lab, closing the gap between a sequence that looks
            good on paper and one a manufacturer can deliver.
          

          
    
            
  • Synthesisability
    • 
            
  • Expression
    • 
            
  • Codon optimisation
    • 
            
  • GC content & repeats
    • 
            
  • Secondary structure
    • 
          

          
    
            
  • —Parameters
    • 
            
  • —Training sequences
    • 
            
  • —Manufacturability lift
    • 
          

          

            View on GitHub →
          

        


        

          

            
Alispec‑1

            In training
          

          
Mass‑spec → structure · analytical methods

          

            Alispec‑1 works the analytical side. Given mass spectrometry data, it
            predicts the molecular structure behind a spectrum — turning raw analytical output into a
            candidate structure. It's the piece that lets the agents reason about what was actually
            measured in the lab, not just what was designed, and close the design‑make‑test
            loop with real data.
          

          
    
            
  • De novo structure elucidation
    • 
            
  • Fragment prediction
    • 
            
  • Spectral matching
    • 
          

          
In development — follow the Substack for progress.

        

      

    


    
    

      

        

          
Writing

          
Notes from the lab that isn't a lab.

        

        All posts on Substack →
      

      

      
    


    
    

      
Collaborate with Aliquant.

      
Wet-lab partners, clinicians working on paediatric enteroviruses, and investors — we'd love to hear from you.

      hello@aliquant.bio