scLAMBDA — hiPSC Perturbation Model (PARK7 example)

📂 Contents

• GSE213069_GPT.h5ad
  Single-cell RNA-seq dataset after filtering, retaining only genes covered by GenePT text embeddings (text-embedding-3-large).
  This ensures full compatibility with scLAMBDA training and inference.
• scLAMBDA_hiPSC_GPT.ipynb
  Full training and inference pipeline (environment setup → model initialization → training → perturbation prediction → visualization).
• embedding_coverage_report.csv
  Report showing which genes in the hiPSC dataset are covered by GenePT embeddings, and the overall coverage rate.
• ckpt.pth
  Checkpoint files from fine-tuning scLAMBDA on the hiPSC dataset.
  If included, they can be directly loaded for downstream prediction.
• GenePT_gene_protein_embedding_model_3_text.pickle
  Pretrained gene/protein embeddings generated by GenePT (dimension = 3072). Used as input embeddings for perturbation modeling and prediction.
• Fast_prediction.ipynb
  Simple prediction script without training.

🧬 Dataset details

• Source: GSE213069 (Perturb-seq, hiPSC)
• Cells: ~25k hiPSC cells (CTRL + 22 perturbations)
• Genes: ~60k original → filtered to ~20k genes with valid embeddings
• Format: .h5ad (AnnData, Scanpy-compatible)

⚙️ Training workflow (summary)

1. Environment
   • GPU: RTX4090 (24GB)
   • PyTorch 2.1.2 (CUDA 11.8)
   • Python 3.10
2. Filter dataset
   python

   1adata = sc.read_h5ad("raw_data.h5ad") 2covered = [g for g in adata.var_names if g in gene_embeddings] 3adata = adata[:, covered].copy() 4adata.write_h5ad("adata_filtered_for_embedding_covered_genes.h5ad")
3. Train scLAMBDA
   python

   1model = sclambda.model.Model( 2 adata, 3 gene_embeddings, 4 training_epochs=500, 5 batch_size=1000, 6 latent_dim=50, 7 hidden_dim=1024, 8 lambda_MI=500, 9 multi_gene=False 10) 11model.train()
4. Predict perturbations (example: PARK7)
   res = model.predict(["PARK7"], return_type="cells")

🔗 References

• scLAMBDA: https://github.com/gefeiwang/scLAMBDA
• Perturb-seq dataset (GSE213069): NCBI GEO
• GenePT embeddings: Zenodo resource

📝 Citation

• Wang et al. scLAMBDA: Language Models for Predicting Gene Perturbation Effects in Single Cells (GitHub: gefeiwang/scLAMBDA)
• Gefei Wang, Tianyu Liu, Jia Zhao, Youshu Cheng, Hongyu Zhao. Modeling and predicting single-cell multi-gene perturbation responses with scLAMBDA. bioRxiv 2024; doi: https://doi.org/10.1101/2024.12.04.626878.