Peptiter / DiscoveryLab
Discovery
Discovery story

Natural seeds — parent peptide library and guardrails.

The natural-seed library and its guardrails — DiscoveryLab generates around evolved peptide families, not from a blank slate.

Natural peptide seed library

Endogenous and animal peptides now seed bounded analog discovery.

DiscoveryLab can start from curated parent peptides such as GLP-1, exendin-4, GIP, PYY, somatostatin, salmon calcitonin, LL-37, hepcidin, and ziconotide lineage scaffolds, then stabilize or evolve variants under explicit evidence guardrails.

16
seed records
13
direct design seeds
3
mechanism-only systems
Open endogenous and animal peptides module
Bounded analog agents
9
seed-scoped agents
108
default hypotheses
54
Pareto retained

The Swift agent runner starts one constrained discovery loop per parent peptide, records ancestry on every candidate, and blocks mechanism-only systems until a curated peptide sequence or motif bridge exists.

Parent peptide pickersource-linked
GLP-1 active core
human gut incretin
GLP1R
stabilize analogs
Exendin-4
Gila monster saliva
GLP1R
evolve long-acting variants
GIP
human enteroendocrine
GIPR
tune incretin selectivity
PYY
human satiety peptide
NPY2R
improve stability
Somatostatin-14
human endocrine brake
SSTR2 / SSTR5
constrain cyclic analogs
Salmon calcitonin
non-human ortholog
CALCR
compare potency liabilities
LL-37
human cathelicidin
AMP
reduce lytic risk
Hepcidin-25
human iron regulator
SLC40A1
improve manufacturability
Ziconotide lineage
cone snail venom
Cav2.2
preserve channel selectivity
Design intents
Stabilize
protease liability, terminal capping, cyclization, solubility
Evolve analogs
bounded mutation around a known parent sequence
Improve selectivity
target receptor retention with off-target checks
Mechanism inspiration
frontier systems stay non-mutational until sequence evidence exists
Evidence guardrails
CHK

Direct mutation is enabled only for seeds with a recorded sequence.

CHK

Dsup, bear hibernation, and antifreeze biology are mechanism inspiration, not drug claims.

CHK

Natural origin never implies safety, dose, efficacy, or clinical readiness.

CHK

Every generated family still needs fit, stability, synthesis, toxicity, and mechanism gates.

Preliminary agent outcomes

The first default run completed all nine seed-scoped agents and retained a six-candidate Pareto panel for each lineage. These are computational hypotheses only; every row still requires folding, receptor-fit, stability, synthesis, and safety evidence before promotion language.

runner
peptiter-natural-analog-agents
settings
2 generations · 6 candidates per seed
checksum
5cc907856c650a7
blocked
0 seed agents
SeedTargetIntentTop candidateFitnessUncertaintyRole
GLP-1 active core
GLP1R
stabilize
CAND-EVO-2-5-0315
0.744
0.244
balanced · liability edit
Exendin-4
GLP1R
evolve
CAND-EVO-1-2-5757
0.649
0.244
balanced · local analog
GIP
GIPR
selectivity
CAND-EVO-1-6-8012
0.685
0.244
target activity · liability edit
PYY
NPY2R
protease
CAND-EVO-2-5-2503
0.739
0.322
balanced · liability edit
Somatostatin-14
SSTR2
selectivity
CAND-EVO-2-5-6318
0.692
0.244
target activity · liability edit
Salmon calcitonin
CALCR
ortholog
CAND-EVO-2-6-5381
0.547
0.236
target activity · local analog
LL-37
AMP
selectivity
CAND-EVO-1-5-0443
0.643
0.338
safety profile · crossover
Hepcidin-25
SLC40A1
manufacture
CAND-EVO-1-5-4419
0.717
0.338
balanced · crossover
Ziconotide lineage
CACNA1B
selectivity
CAND-EVO-2-4-7521
0.670
0.260
target activity · crossover