AI Computational Chemistry

Rank your compound series before you synthesize any of them.

Alkira scores each candidate on binding affinity, toxicity risk, and synthesizability, returning a ranked shortlist so bench chemists work only on molecules that clear all three filters.

Used by research chemists at 8 biotech teams
Compound Binding Tox Risk Synth % Rank
KB-4821 8.2 Low 85% 1
AL-0934 7.9 Low 78% 2
MB-2217 8.5 Medium 54% 3
TR-6688 6.8 Low 92% 4
CK-1055 7.1 High 71% filtered

The bottleneck

Synthesis time is the scarcest resource in early discovery.

Medicinal chemists spend weeks synthesizing a series only to find that most candidates fail the same predictable filters. Binding pocket geometry, metabolic liability, and step count are computable before any molecule touches a flask.

Alkira runs those filters computationally first, so the bench queue contains only compounds that have already cleared the most common failure modes.

Synthesis is expensive and slow

A single compound can take 2 to 6 weeks of bench time. Without pre-screening, most of that effort goes toward molecules that fail predictable filters.

Series assessment is manual and slow

Chemists assess dozens of candidates using separate docking tools, tox models, and retrosynthesis software. Combining scores into a ranked shortlist takes days.

Bad picks erode program momentum

Late-stage failures compound. Each avoidable failure delays the program and consumes budget that could support additional series exploration.

Three-filter scoring

One upload. Three scores. A ranked shortlist in 90 seconds.

Upload your SMILES file and Alkira runs binding, tox, and synthesizability in parallel. The output is a ranked table your team can act on immediately.

01

Binding affinity scoring

Structure-based docking against your specified target pocket, returning predicted binding free energy for each candidate in the series.

02

Toxicity risk prediction

Multi-model ensemble trained on public ADMET datasets flags hERG liability, metabolic instability, and reactive group alerts, classifying each compound Low, Medium, or High.

03

Synthesizability estimation

Retrosynthesis-based scoring estimates achievable synthesis routes and reports a percent score reflecting ease of preparation from commercially available building blocks.

Research outcomes

Numbers from the bench teams using Alkira today

These figures reflect tool-workflow performance measured across research teams actively using Alkira on their compound series. Individual results vary by program.

8 research teams actively using Alkira on live drug programs
90 sec median scoring run time from SMILES upload to ranked output
62% median reduction in synthesis queue after Alkira pre-screening

"We ran a 40-compound series through Alkira before committing to synthesis. The scoring cut our queue to 15 molecules and the two flagged for high tox would have been our first two failures. That is three weeks of bench time recovered."

Medicinal chemist Small-molecule oncology program, Melbourne

"The synthesizability scores align well with what we see on the bench. When Alkira gives something above 80 percent, it reliably comes together in four to six steps. Below 50 it is usually a multi-week project. That calibration alone is valuable."

Senior research chemist Early-stage biotech, Sydney

Pricing

Straightforward pricing for individual researchers and teams

Start with a free ranking run. No credit card required.

Researcher
$199 /mo

Up to 500 compound submissions per month. Full three-filter scoring, CSV export, and run history.

Start free trial

See full pricing including Enterprise

Stop synthesizing molecules that were always going to fail.

Submit your compound series today and get a ranked shortlist in 90 seconds. No installation, no contract, one free run to start.