Pheno
PlatformSequence & Structure

Computational biology tools, through your LLM.

Search, fold and compare, inside one investigation.

Strain questions turn into protein questions fast. Pheno answers them inside the same investigation, so the search, the model and the comparison land in the same report as the route.

FTLGRHQFENMIFCVKHNLRGDSLASEQFPHQRQRHEDGCQFDGHDTAIMKSSRMPTCFLEHDC
Structure

Fetch a model, or fold one.

AlphaFold models when they exist. A prediction when they don't.

Existing models are checked against your sequence, and a prediction runs on an isolated worker when none exists. Confidence is reported per residue and measures the model’s certainty, never activity.

Structure
very high confident low very low
  1. Queued
  2. Folding
  3. Scoring
  4. Model ready
Mean confidence
84
Residues
320
Low-confidence stretches
4
Source
new prediction
The backbone is generated for the figure, not a real protein. Confidence bands follow the model's own per-residue score.
Comparison

Host enzyme, meet donor.

A structural overlay with the numbers that matter.

Overlap, coverage, deviation and a downloadable overlay. A matching fold is a reason to look closer, and an assay decides.

Structure comparison
loop differshostdonor
Overlap score
0.82
normalised to the host; 1 is identical
Deviation
1.9 Å
over the aligned backbone
Aligned
94%
of host residues
Reading
similar fold; one loop differs near the pocket
Scores are from a structural aligner over saved models: overlap normalised to each protein, aligned coverage and deviation, with a backbone overlay you can download. A similar fold is a place to look, not a working enzyme. Backbones are generated.
Evidence

Every result stays attached to the question.

Every search, model and comparison is saved with its inputs and linked from the report that used it.

What each tool promises
Host search
local, private, hits mapped to genes
Public search
opt-in, queued, rate-limited
Model lookup
existing model only, sequence checked
Prediction
isolated worker, cancellable, progress visible
Comparison
overlap, coverage, deviation, overlay file
Confidence by residue
7062–84192–206302–338398–411residue 1412
A per-residue score from the model itself. It measures how sure the model is about local geometry, not whether the protein is active. Values are generated.
Next