Repository navigation
PD-14: Add 10.x pipeline notebook series - #16
Merged
Merged
Conversation
Seven new notebooks demonstrating the biolmai.pipeline (DataPipeline / DuckDBDataStore) SDK, revised to match repo conventions: standard BioLM HTML headers and footers, 10.x series numbering, and typo fixes. Also adds biolmai[pipeline] to requirements.txt. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The biolmai.pipeline module is not yet publicly released. Adds a blockquote notice to each of the 10.x notebooks directing users to contact BioLM for early access.
- Remove⚠️ Preview Feature banner from all 7 notebooks (pipelines is GA) - Change `biolmai[pipeline]` install to `biolmai` (now ships with full package) - Update titles to be more descriptive and biology-campaign-focused: 10.1: Skip Repeated API Calls: Cache and Reuse Predictions Across Screening Runs 10.2: Select a Diverse Peptide Library by Clustering ESM2 Embeddings 10.3: Run a Screening Campaign Across Multiple Batches Without Repeating Predictions 10.4: Scan All Single-Point Variants for Thermal Stability and Solubility 10.5: Explore Protein Sequence Space: Similarity, Outliers, and Property Landscapes 10.6: Filter Designs by Fold Quality Before Running Stability and Solubility Predictions - Execute 10.0–10.3, 10.5 to add real API outputs (DataFrames, plots, pipeline logs) - Fix 10.4: add ValidAminoAcidFilter, make tm/sol stages parallel (depends_on=["validate"]), improve wt_tm lookup error message; outputs pending API recovery - Ignore .biolm/ pipeline cache dir and *.duckdb files
- Replace temperature-regression with temberture-regression (correct v3 slug) in all 10.x notebooks - Execute 10.4 and 10.6 with live API to embed real outputs - Fix 10.4 ThresholdFilter threshold (55°C → 48°C, matched to WT Tm of 51°C) - Replace boltz2 with ESMFold in 10.6 (boltz2 requires custom molecules input format not supported by the pipeline SDK; ESMFold returns identical mean_plddt/ptm metrics and accepts plain sequence strings) - Fix 10.4 source encoding corruption from previous session (char-split source repaired)
- Use add_cofolding_prediction (CofoldingPredictionStage) for correct molecules format required by boltz2 - Extract confidence.complex_plddt and confidence.ptm as filterable scalar predictions via new SDK extractions support - Fix plddt ThresholdFilter threshold from 65.0 to 0.65 (boltz2 returns values in [0,1] scale, not [0,100]) - Fix ESMFold references in markdown header and note - All three gates execute: fold (13→12) → tm (12→11) → solubility top-5
…boltz2) Folding is more expensive than stability/solubility predictions. The previous order (boltz2 first) was backwards — spend the costly structural validation budget only on sequences that survive the cheap filters. New order: temberture-regression → biolmsol top-7 → boltz2 validate survivors. Boltz-2 now runs on 7 sequences instead of 13.
New titles (agreed): launch campaign / persist predictions / cover more space / trickle pattern / predict mutations / embedding navigation / properties before structure. Removed pip install code fence from requirements cell across all seven notebooks; kept the BIOLMAI_TOKEN export line as plain inline code.
ChanceChallacombe
force-pushed
the
pd-14-pipeline-notebooks
branch
from
July 15, 2026 17:23
f99a7aa to
441291e
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
DataPipelineSDK with DuckDB cachingNotebooks
Other changes
.gitignore: added.biolm/and*.duckdb(SDK cache files)requirements.txt: addedbiolmai[pipeline]Test plan
BIOLMAI_TOKENos.environ.get).duckdb,.biolm/)