Your mouse antibody, made human.
Self-serve antibody humanization, benchmarked against the Observed Antibody Space. What used to be a six-figure CRO engagement, now a $200/month platform.
Built on peer-reviewed open source
- OASis
- BioPhi
- Sapiens
- ANARCI
- ImmuneBuilder
- Boltz-2
The algorithms haven’t changed. The access has.
The same peer-reviewed methods CROs charge six figures for are open source. We built the pipeline around them so you don’t have to.
The CRO route
- 1 weekSend sequences. Wait for a quote.
- 1 weekReview quote. Sign the SOW.
- 6–12 weeksWait for results.
- $50–150KPer project. No iterations included.
With κλllima
- NowPaste your VH and VL sequences.
- < 5 minANARCI, CDR grafting, Sapiens, OASis — done.
- NowCompare variants. Flag liabilities. Export.
- $200/moIterate as many times as your credits allow.
Three steps
How it works
Paste your sequences
VH and VL amino acid sequences, raw or FASTA. Or upload a CSV to run your whole panel at once.
Run the pipeline
Choose humanization, structure prediction, immunogenicity, or the full stack. Jobs run in parallel in the background.
Compare and export
Side-by-side variant comparison, per-residue scoring, liability flags. Export CSV, multi-FASTA, or PDF.
Verified benchmark
We ran the trastuzumab precursor. You can check our work.
muMAb4D5 is the murine antibody Carter et al. humanized at Genentech in 1992 to create trastuzumab. We ran it through κλllima and compared.
muMAb4D5 — murine parent
OASis humanness — VH
0.366
OASis humanness — VL
0.454
Observed Antibody Space mean score
After κλllima — CDR grafting, IGHV3-23 + IGKV1
OASis humanness — VH
0.616
+68%
OASis humanness — VL
0.788
+73% · 23 framework substitutions
Compared to Carter et al. 1992: κλllima’s CDR grafting on IGHV3-23 + IGKV1 reproduces the framework substitution choices made to develop trastuzumab (huMAb4D5-5). One residue differs at VH position 5. Run the same murine sequences yourself to verify.
That was 1992. This is 2026.
The same humanization. Plus everything they didn’t have.
A κλllima run reproduces Carter’s framework choices — and delivers five additional analyses in the same session that didn’t exist in 1992.
01
Sequence alignment
muMAb4D5 parent vs. humanized — framework substitutions in red/green, CDRs preserved
02
OASis humanness scoring
Observed Antibody Space percentile score before and after CDR grafting
OASis humanness — Observed Antibody Space
VH
VL
03
MHC-I/II immunogenicity
T-cell epitope risk scored per peptide across 70+ HLA alleles
MHC-I/II immunogenicity
GLEWIGRIYPT
HLA-DRB1*07:01
KASQDVNTAVA
HLA-A*02:01
IYPTNGYTRY
HLA-DRB1*15:01
SLQPEDFATYY
HLA-DRB1*04:01
04
Sequence liabilities
Deamidation, oxidation, and isomerization motifs flagged by position
Sequence liabilities
VH-N28
Deamidation · NG
VH-W36
Oxidation · W
VL-N30
Deamidation · NS
VL-T89
Isomerization · DT
VH-M107
Oxidation · M
05
3D structure — ImmuneBuilder
Humanized trastuzumab (4D5) VH/VL structure predicted by ImmuneBuilder, rendered live in Molstar
Pricing
One credit currency. Pick the tier that fits your month.
Every plan refreshes a monthly credit grant. Spend credits on any pipeline your plan unlocks. Run out? Buy a credit pack that never expires.
Free — $0 / month
30 credits each month. Try the humanization pipeline end-to-end, no credit card.
Team — $580 / seat / month · min. 3 seats
Everything in Advance for each member — including API access — plus shared org, admin controls, and custom subdomain.
Academic — free
Full Advance-level access, auto-approved for .edu and other institutional email domains. Buy credit packs as needed.
Frequently asked questions
What input format do you accept?
Amino acid sequences for VH and optionally VL chains. Paste raw sequence, FASTA format, or upload a CSV for batch processing.
How does pricing work?
Each plan includes a monthly credit grant. Every action (humanization, structure, stability, immunogenicity, complex prediction) costs a fixed number of credits. Out of credits? Top up with a pack that never expires. Failed jobs are refunded automatically.
Is my data secure?
Your sequences are stored in a row-isolated PostgreSQL database and are never sold or shared. Paid plans opt out of training data use by default; Free plans can opt out in Settings → Privacy. We're working toward SOC 2 compliance.
What species do you support?
Any species origin — the humanization pipeline targets human frameworks regardless of the input species. Mouse, rat, rabbit, and llama/camelid sequences are all supported.
Your first humanization takes four minutes.
30 free credits every month, no card required. Enough to humanize a candidate, run OASis scoring, and flag the liabilities before your CRO quote even arrives.