novadde

Plugin novadde from marketplace novadde-plugin in geodesicintelligence/novadde-plugin. Commands read from the repo's manifest — 1 month ago.

Skills (33)

• antibody-developability-thresholds

  Use when a developability number needs a threshold — TAP's five descriptors (total CDR length, PSH, PPC, PNC, SFvCSP), where the amber and red lines actually sit, and whether a given flag transfers to your molecule. This is the threshold half of `antibody-liabilities`, which computes sequence motifs

• antibody-interface-metrics

  Use when reading or ranking the numbers a co-folding model returns for an antibody-antigen complex — ipTM, ipSAE, pLDDT, PAE, a ranking score, an ESM-2 likelihood, a contact fraction — or when setting a gate on them. Covers why ipTM flatters a small paratope, what ipSAE changes, which distance cutof

• antibody-liabilities

  Use when an antibody or nanobody sequence needs the liability and charge screen actually run rather than described — deamidation, isomerisation, Asp-Pro, N-glycosylation sequon, unpaired cysteine, pI, hydrophobic runs. This is the executable half of `antibody-interface-metrics`, which owns what the

• antibody-numbering

  Use when antibody residue numbering, CDR definitions, or chain conventions matter — designing CDRs, calling rfantibody/boltzgen, profiling CDR lengths, or interpreting a CDR length that came back as 0. Triggers include "which CDR residues", "Chothia or IMGT", "what numbering does this model want", "

• binder-benchmarking

  Benchmark a generative binder/peptide design model (BoltzGen and similar tools on geodesic) against targets with an experimentally validated ligand before trusting it on a real campaign. Use when standing up or re-validating a design pipeline, when hit rate looks target-dependent and it is unclear w

• binder-fold-stability

  Use when a designed binder scores well in complex and you need to know whether it is actually a folded, on-target molecule — the apo/holo check. Covers folding the binder alone and measuring apo↔holo Cα RMSD, reading apo pLDDT without falling for the scale and B-factor traps, and measuring hotspot c

• biopython

  Comprehensive molecular biology toolkit. Use for sequence manipulation, file parsing (FASTA/GenBank/PDB), phylogenetics, and programmatic NCBI/PubMed access (Bio.Entrez). Best for batch processing, custom bioinformatics pipelines, BLAST automation. For quick lookups use gget; for multi-service integ

• bioservices

  Unified Python interface to 40+ bioinformatics services. Use when querying multiple databases (UniProt, KEGG, ChEMBL, Reactome) in a single workflow with consistent API. Best for cross-database analysis, ID mapping across services. For quick single-database lookups use gget; for sequence/file manipu

• borzoi

  Predict genome-wide functional tracks (RNA-seq, CAGE, DNase, ChIP) from DNA sequence with Borzoi. Use this skill when: (1) Scoring the regulatory effect of a variant on expression/accessibility, (2) Generating predicted coverage tracks for a locus, (3) Prioritising non-coding variants by predicted t

• environment-orientation

  Use FIRST, before concluding that a tool or package is missing, and whenever something needs installing. Says where this container actually keeps things — which interpreter `python3` is, what is already importable, which bioinformatics binaries are on PATH, which skills directory is really read, and

• epitope-selection

  Use when choosing WHERE on a target a binder or antibody should bind — picking hotspot residues, reading UniProt site annotations, or deciding whether a proposed epitope is reachable at all. Every other design skill here begins after an epitope exists; this one is about choosing it, and about the an

• evo2

  Score, embed, and generate DNA sequences with Evo 2, a long-context genomic foundation model. Use this skill when: (1) Computing per-nucleotide or per-sequence likelihoods for variant effect scoring, (2) Embedding genomic windows for downstream classification, (3) Generating DNA conditioned on a pre

• fair-esm2

  Embed proteins with Meta AI's ESM-2 (`fair-esm` package). Use this skill when: (1) Extracting per-residue or per-sequence embeddings for downstream ML, (2) Masked-LM likelihood / mutation effect scoring, (3) Contact prediction from a sequence. vendored-from: https://github.com/HughYau/AcademicForge

• geodesic

  Run protein design, structure prediction and structure/sequence validation on this deployment's own GPUs through the geodesic MCP server. Use for de novo binder design, antibody and nanobody CDR design, structure prediction of complexes and ligands, inverse folding, all-atom structure validation, an

• gget

  Fast CLI/Python queries to 20+ bioinformatics databases. Use for quick lookups: gene info, BLAST/BLAT, viral sequence downloads, AlphaFold structures, enrichment analysis, OpenTargets, COSMIC, CELLxGENE, and 8cube mouse specificity/expression data. Best for interactive exploration and simple queries

• glycoengineering

  Analyze and engineer protein glycosylation. Scan sequences for N-glycosylation sequons (N-X-S/T), predict O-glycosylation hotspots, and access curated glycoengineering tools (NetOGlyc, GlycoShield, GlycoWorkbench). For glycoprotein engineering, therapeutic antibody optimization, and vaccine design.

• ipsae-calculate

  Use when ipSAE or interface PAE has to be computed from a PAE matrix and a structure, rather than read off whatever a folding model already printed. Takes any PAE matrix plus the matching mmCIF/PDB, including this deployment's own predictors. This is the executable half of `antibody-interface-metric

• kernel-background-processes

  Use before starting a server, watcher, build, download or test run that may outlast one command. The container is stopped after an idle timeout (10 min by default) and a background job does not count; /tmp is a small RAM disk; port 8000 is this session's own server. Coding work, not OS kernel proces

• kernel-git-bisect

  Use when something that used to work now fails and several commits lie between - a regression, "it worked last week", "which commit broke this". Find the first bad commit with git bisect run and a time-bounded check script (exit 125 skips a commit), then git bisect reset. Coding work, not an OS or J

• kernel-git-identity-and-commits

  Use before the first git commit in a repo, or on "Author identity unknown" or "Please tell me who you are". No git identity is set up here - ask the user for the name and email, set them for this repo only, never invent one. Commit with -m, stage by path, never force-push unless asked. Coding work,

• kernel-merge-conflicts

  Use when git reports CONFLICT, a merge, rebase, cherry-pick or stash pop stops, or a file holds <<<<<<< markers. Read both sides and the base, resolve by intent, run the tests, then finish without an editor (GIT_EDITOR=true git rebase --continue, git commit --no-edit). Coding work, not an OS or Jupy

• kernel-test-first

  Use when fixing a bug or adding a regression test or guard. Write the test first and watch it fail for the reported reason, fix the code, then put the bug back on purpose and watch the test go red again - a test that also passes against the bug guards nothing. Coding work, not an OS or Jupyter kerne

• kernel-verify-before-done

  Use before saying done, fixed, passing or works. Find the project's own test runner (pytest may not be installed), run the command that proves the claim, and read its exit code and output. Zero tests collected, all skipped, or a pipe that hides a failure is not evidence. Coding work, not an OS or Ju

• molecular-dynamics

  Run and analyze molecular dynamics simulations with OpenMM and MDAnalysis. Set up protein/small molecule systems, define force fields, run energy minimization and production MD, analyze trajectories (RMSD, RMSF, contact maps, free energy surfaces). For structural biology, drug binding, and biophysic

• paper-lookup

  Search 11 academic literature APIs for papers, preprints, citations, and open-access full text, and return results with reproducible provenance. Covers PubMed, PMC (full text), Europe PMC (full-text and preprint search), bioRxiv, medRxiv, arXiv, OpenAlex, Crossref, Semantic Scholar, CORE, Unpaywall.

• plm-variant-scoring

  Use when scoring or ranking protein variants with a language model likelihood — which convention to use (masked-marginal, wt-marginal, mutant-marginal, pseudo-log-likelihood), what each costs, which direction the sign runs, why a multi-mutant needs a different forward pass than a point mutant, and w

• protein-binder-campaign

  Run a de novo miniprotein binder design campaign following the Anthropic claude-protein-binder-design release prompts. Use when asked to design binders against a protein target end to end, to reproduce or adapt that campaign's methodology, or when questions come up about its roster, gates, scoring i

• rdkit

  Cheminformatics toolkit for fine-grained molecular control. SMILES/SDF parsing, descriptors (MW, LogP, TPSA), fingerprints, substructure search, 2D/3D generation, similarity, reactions. For standard workflows with simpler interface, use datamol (wrapper around RDKit). Use rdkit for advanced control,

• scgpt

  Embed and annotate single-cell expression data with scGPT, a foundation model for single-cell biology. Use this skill when: (1) Producing cell embeddings from an AnnData for clustering/integration, (2) Zero-shot or fine-tuned cell-type annotation, (3) Gene-level representation for perturbation/GRN t

• scientific-visualization

  Create and audit truthful, accessible, publication-ready scientific figures with Matplotlib, Seaborn, or Plotly. Use for figure design, multi-panel layouts, uncertainty and missing-data displays, color/contrast review, image metadata validation, and journal export planning.

• scientific-writing

  Draft, revise, and audit scientific manuscripts or reports with explicit evidence provenance, reporting-guideline coverage, authorship accountability, confidentiality controls, and local consistency checks. Use for manuscript sections, references, declarations, tables, figures, or submission prepara

• setup

  Connect the installed Novadde plugin to the production Model Platform through shared OAuth, verify its account, reconnect, or select legacy API-key mode explicitly.

• status

  Report how this session is connected to the Geodesic Model Platform - credential, the account's allowance (daily submissions or weekly credits), recent jobs, and whether the background job watcher can see them. Use before submitting work, when the user asks what is running, or when a job tool behave

Subagents (1)

• novadde

  NovaDDE, Geodesic Intelligence's AI co-scientist for drug discovery. Designs peptides, antibodies and protein binders by running the lab's own GPU models, confirms every submission with the user first, and keeps evidence tiers apart. Built from the deployment's own prompt.

MCP servers (1)

• model_platform

  connects to: https://platform.geodesiclab.com/api/mcp

Hooks (4)

• SessionStart

  ${CLAUDE_PLUGIN_ROOT}/scripts/session-brief.sh

• UserPromptSubmit

  ${CLAUDE_PLUGIN_ROOT}/scripts/user-suffix.sh

• PreToolUse

  ${CLAUDE_PLUGIN_ROOT}/scripts/submit-guard.sh

• PostToolUse

  ${CLAUDE_PLUGIN_ROOT}/scripts/record-job.sh

Similar plugins