Contact
ContactClient login

Eliminate the bottlenecks in drug discovery and development

Alosenta finds promising new drug candidates and antibodies, identifies and eliminates high-risk candidates before they cost a preclinical trial, picks the best drug combinations, and puts a price on every risk, including wet lab validation.

Get in contactRead more

Six modules. One platform.

90%

of drug candidates fail in clinical trials.

Sun et al., Acta Pharm Sin B, 2022
30%

of those failures are toxicity. The molecule harms the patient.

Sun et al., Acta Pharm Sin B, 2022

Caught before the preclinical.

ToxSMILES flags the structures known to cause harm. Galaxy maps the human proteins a molecule may hit. The wet lab confirms the high-risk ones and FinSMILES puts a price on each risk, so you drop them before you pay for the study.

Galaxy finds 96.4% of a new molecule’s real targets in its top 20, on molecules it had never seen.

Many cancers are treated with drugs in combination.

Multiple myeloma, a blood cancer, is treated with three or four drugs together. Which ones to put together is the hard part: every drug you add multiplies the options to test.

NCCN Guidelines, Multiple Myeloma
406

pairs and triplets to choose from, for just 16 myeloma drugs.

Alosenta, multiple myeloma study, 2026
218

ruled out by Combiner, because the drugs in them do the same job twice or their side effects stack up.

Alosenta, multiple myeloma study, 2026

It found the treatments doctors already use.

Nobody told Combiner which combinations work. Two of its top five are today’s standard myeloma treatments, and a third is in a phase 3 trial. It took 29 minutes.

  1. Dexamethasone + Bortezomib + IberdomideIn phase 3
  2. Dexamethasone + Bortezomib + ThalidomideStandard of care, VTd
  3. Dexamethasone + Bortezomib + LenalidomideStandard of care, VRd
  4. Dexamethasone + Bortezomib + PomalidomideEstablished regimen
  5. Selinexor + Bortezomib + IberdomideNew, synergy explained

Every new drug starts from a hit.

A hit is a molecule that already acts on the target, but rarely well enough or safely enough to be a drug. Chemists improve it one change at a time.

10⁶⁰

drug-like molecules could exist. No lab can try them all.

Bohacek et al., Med Res Rev, 1996

DrugForge searches it from your hit.

It proposes variants that keep the part that works, meet the properties you set, and come with a score for how hard each is to make.

Then it goes again.

The best variant becomes the next starting point, round after round, and each round is screened by ToxSMILES and Galaxy before anything is made. Interfacer does the same for antibodies, from 190,000 experimental structures.

Why now

AI now designs molecules faster than labs can test them. The new bottleneck is knowing which ones are safe.

  1. The FDA publishes its roadmap to replace animal testing in safety studies, computer models included.

    Source

  2. NIH stops funding research that relies on animal models alone.

    Source

  3. FDA draft guidance on new approach methodologies in drug development.

    Source

  4. FDA rules now count computer models and human cell tests as nonclinical tests.

    “methods using human cells, organs-on-chips, computer models and other advanced technologies”
    Source

Start with one pilot

One scope, one fixed price, your molecules. You see the answer before you commit to more.

Get in contact