@bgicli/bgicli 2.1.1 → 2.2.1

package/data/skills/bulk-combat-correction/SKILL.md

@@ -0,0 +1,54 @@
+---
+name: bulk-rna-seq-batch-correction-with-combat
+title: Bulk RNA-seq batch correction with ComBat
+description: Use omicverse's pyComBat wrapper to remove batch effects from merged bulk RNA-seq or microarray cohorts, export corrected matrices, and benchmark pre/post correction visualisations.
+---
+
+# Bulk RNA-seq batch correction with ComBat
+
+## Overview
+Apply this skill when a user has multiple bulk expression matrices measured across different batches and needs to harmonise them
+ before downstream analysis. It follows [`t_bulk_combat.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_bulk_combat.ipynb), w
+hich demonstrates the pyComBat workflow on ovarian cancer microarray cohorts.
+
+## Instructions
+1. **Import core libraries**
+   - Load `omicverse as ov`, `anndata`, `pandas as pd`, and `matplotlib.pyplot as plt`.
+   - Call `ov.ov_plot_set()` (aliased `ov.plot_set()` in some releases) to align figures with omicverse styling.
+2. **Load each batch separately**
+   - Read the prepared pickled matrices (or user-provided expression tables) with `pd.read_pickle(...)`/`pd.read_csv(...)`.
+   - Transpose to gene × sample before wrapping them in `anndata.AnnData` objects so `adata.obs` stores sample metadata.
+   - Assign a `batch` column for every cohort (`adata.obs['batch'] = '1'`, `'2'`, ...). Encourage descriptive labels when availa
+ble.
+3. **Concatenate on shared genes**
+   - Use `anndata.concat([adata1, adata2, adata3], merge='same')` to retain the intersection of genes across batches.
+   - Confirm the combined `adata` reports balanced sample counts per batch; if not, prompt users to re-check inputs.
+4. **Run ComBat batch correction**
+   - Execute `ov.bulk.batch_correction(adata, batch_key='batch')`.
+   - Explain that corrected values are stored in `adata.layers['batch_correction']` while the original counts remain in `adata.X`.
+5. **Export corrected and raw matrices**
+   - Obtain DataFrames via `adata.to_df().T` (raw) and `adata.to_df(layer='batch_correction').T` (corrected).
+   - Encourage saving both tables (`.to_csv(...)`) plus the harmonised AnnData (`adata.write_h5ad('adata_batch.h5ad', compressio
+n='gzip')`).
+6. **Benchmark the correction**
+   - For per-sample variance checks, draw before/after boxplots and recolour boxes using `ov.utils.red_color`, `blue_color`, `gree
+n_color` palettes to match batches.
+   - Copy raw counts to a named layer with `adata.layers['raw'] = adata.X.copy()` before PCA.
+   - Run `ov.pp.pca(adata, layer='raw', n_pcs=50)` and `ov.pp.pca(adata, layer='batch_correction', n_pcs=50)`.
+   - Visualise embeddings with `ov.utils.embedding(..., basis='raw|original|X_pca', color='batch', frameon='small')` and repeat fo
+r the corrected layer to verify mixing.
+7. **Troubleshooting tips**
+   - Mismatched gene identifiers cause dropped features—remind users to harmonise feature names (e.g., gene symbols) before conca
+tenation.
+   - pyComBat expects log-scale intensities or similarly distributed counts; recommend log-transforming strongly skewed matrices.
+   - If `batch_correction` layer is missing, ensure the `batch_key` matches the column name in `adata.obs`.
+
+## Examples
+- "Combine three GEO ovarian cohorts, run ComBat, and export both the raw and corrected CSV matrices."
+- "Plot PCA embeddings before and after batch correction to confirm that batches 1–3 overlap."
+- "Save the harmonised AnnData file so I can reload it later for downstream DEG analysis."
+
+## References
+- Tutorial notebook: [`t_bulk_combat.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_bulk_combat.ipynb)
+- Example inputs: [`omicverse_guide/docs/Tutorials-bulk/data/combat/`](../../omicverse_guide/docs/Tutorials-bulk/data/combat/)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-deg-analysis/SKILL.md

@@ -0,0 +1,61 @@
+---
+name: bulk-rna-seq-differential-expression-with-omicverse
+title: Bulk RNA-seq differential expression with omicverse
+description: Guide Claude through omicverse's bulk RNA-seq DEG pipeline, from gene ID mapping and DESeq2 normalization to statistical testing, visualization, and pathway enrichment. Use when a user has bulk count matrices and needs differential expression analysis in omicverse.
+---
+
+# Bulk RNA-seq differential expression with omicverse
+
+## Overview
+Follow this skill to run the end-to-end differential expression (DEG) workflow showcased in [`t_deg.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_deg.ipynb). It assumes the user provides a raw gene-level count matrix (e.g., from featureCounts) and wants to analyse bulk RNA-seq cohorts inside omicverse.
+
+## Instructions
+1. **Set up the session**
+   - Import `omicverse as ov`, `scanpy as sc`, and `matplotlib.pyplot as plt`.
+   - Call `ov.plot_set()` so downstream plots adopt omicverse styling.
+2. **Prepare ID mapping assets**
+   - When gene IDs must be converted to gene symbols, instruct the user to download mapping pairs via `ov.utils.download_geneid_annotation_pair()` and store them under `genesets/`.
+   - Mention the available prebuilt genomes (T2T-CHM13, GRCh38, GRCh37, GRCm39, danRer7, danRer11) and that users can generate their own mapping from GTF files if needed.
+3. **Load the raw counts**
+   - Read tab-delimited featureCounts output with `ov.pd.read_csv(..., sep='\t', header=1, index_col=0)`.
+   - Strip trailing `.bam` segments from column names using list comprehension so sample IDs are clean.
+4. **Map gene identifiers**
+   - Run `ov.bulk.Matrix_ID_mapping(counts_df, 'genesets/pair_<GENOME>.tsv')` to replace `gene_id` entries with gene symbols.
+5. **Initialise the DEG object**
+   - Create `dds = ov.bulk.pyDEG(mapped_counts)`.
+   - Handle duplicate gene symbols with `dds.drop_duplicates_index()` to keep the highest expressed version.
+6. **Normalise and estimate size factors**
+   - Execute `dds.normalize()` to calculate DESeq2 size factors, correcting for library size and batch differences.
+7. **Run differential testing**
+   - Collect treatment and control replicate labels into lists.
+   - Call `dds.deg_analysis(treatment_groups, control_groups, method='ttest')` for the default Welch t-test.
+   - Offer optional alternatives: `method='edgepy'` for edgeR-like tests and `method='limma'` for limma-style modelling.
+8. **Filter and threshold results**
+   - Note that lowly expressed genes are retained by default; filter using `dds.result.loc[dds.result['log2(BaseMean)'] > 1]` when needed.
+   - Set dynamic fold-change and significance cutoffs via `dds.foldchange_set(fc_threshold=-1, pval_threshold=0.05, logp_max=6)` (`fc_threshold=-1` auto-selects based on log2FC distribution).
+9. **Visualise differential expression**
+   - Produce volcano plots with `dds.plot_volcano(title=..., figsize=..., plot_genes=... or plot_genes_num=...)` to highlight key genes.
+   - Generate per-gene boxplots using `dds.plot_boxplot(genes=[...], treatment_groups=..., control_groups=..., figsize=..., legend_bbox=...)`; adjust y-axis tick labels if required.
+10. **Perform pathway enrichment (optional)**
+    - Download curated pathway libraries through `ov.utils.download_pathway_database()`.
+    - Load genesets with `ov.utils.geneset_prepare(<path>, organism='Mouse'|'Human'|...)`.
+    - Build the DEG gene list from `dds.result.loc[dds.result['sig'] != 'normal'].index`.
+    - Run enrichment with `ov.bulk.geneset_enrichment(gene_list=deg_genes, pathways_dict=..., pvalue_type='auto', organism=...)`. Encourage users without internet access to provide a `background` gene list.
+    - Visualise single-library results via `ov.bulk.geneset_plot(...)` and combine multiple ontologies using `ov.bulk.geneset_plot_multi(enr_dict, colors_dict, num=...)`.
+11. **Document outputs**
+    - Suggest exporting `dds.result` and enrichment tables to CSV for downstream reporting.
+    - Encourage users to save figures generated by matplotlib (`plt.savefig(...)`) when running outside notebooks.
+12. **Troubleshooting tips**
+    - Ensure sample labels in `treatment_groups`/`control_groups` exactly match column names post-cleanup.
+    - Verify required packages (`omicverse`, `pyComplexHeatmap`, `gseapy`) are installed for enrichment visualisations.
+    - Remind users that internet access is required the first time they download gene mappings or pathway databases.
+
+## Examples
+- "I have a featureCounts matrix for mouse tumour samples—normalize it with DESeq2, run t-test DEG, and highlight the top 8 genes in a volcano plot."
+- "Use omicverse to compute edgeR-style differential expression between treated and control replicates, then run GO enrichment on significant genes."
+- "Guide me through converting Ensembl IDs to symbols, performing limma DEG, and plotting boxplots for Krtap9-5 and Lef1."
+
+## References
+- Detailed walkthrough notebook: [`t_deg.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_deg.ipynb)
+- Sample count matrix for testing: [`sample/counts.txt`](../../sample/counts.txt)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-deseq2-analysis/SKILL.md

@@ -0,0 +1,50 @@
+---
+name: bulk-rna-seq-deseq2-analysis-with-omicverse
+title: Bulk RNA-seq DESeq2 analysis with omicverse
+description: Walk Claude through PyDESeq2-based differential expression, including ID mapping, DE testing, fold-change thresholding, and enrichment visualisation.
+---
+
+# Bulk RNA-seq DESeq2 analysis with omicverse
+
+## Overview
+Use this skill when a user wants to reproduce the DESeq2 workflow showcased in [`t_deseq2.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_deseq2.ipynb). It covers loading raw featureCounts matrices, mapping Ensembl IDs to symbols, running PyDESeq2 via `ov.bulk.pyDEG`, and exploring downstream enrichment plots.
+
+## Instructions
+1. **Import and format the expression matrix**
+   - Call `import omicverse as ov` and `ov.utils.ov_plot_set()` to standardise visuals.
+   - Read tab-separated count data from featureCounts using `ov.utils.read(..., index_col=0, header=1)`.
+   - Strip trailing `.bam` from column names with `[c.split('/')[-1].replace('.bam', '') for c in data.columns]`.
+2. **Map gene identifiers**
+   - Ensure the appropriate mapping pair exists by running `ov.utils.download_geneid_annotation_pair()`.
+   - Replace `gene_id` with gene symbols using `ov.bulk.Matrix_ID_mapping(data, 'genesets/pair_<GENOME>.tsv')`.
+3. **Initialise the DEG object**
+   - Create `dds = ov.bulk.pyDEG(data)` from the mapped counts.
+   - Resolve duplicate gene names with `dds.drop_duplicates_index()` and confirm success in logs.
+4. **Define contrasts and run DESeq2**
+   - Collect sample labels into `treatment_groups` and `control_groups` lists that match column names exactly.
+   - Execute `dds.deg_analysis(treatment_groups, control_groups, method='DEseq2')` to invoke PyDESeq2.
+5. **Filter and tune thresholds**
+   - Inspect result shape (`dds.result.shape`) and optionally filter low-expression genes, e.g. `dds.result.loc[dds.result['log2(BaseMean)'] > 1]`.
+   - Set thresholds via `dds.foldchange_set(fc_threshold=-1, pval_threshold=0.05, logp_max=6)` to auto-pick fold-change cutoffs.
+6. **Visualise differential genes**
+   - Draw volcano plots with `dds.plot_volcano(...)` and summarise key genes.
+   - Produce per-gene boxplots: `dds.plot_boxplot(genes=[...], treatment_groups=..., control_groups=..., figsize=(2, 3))`.
+7. **Run enrichment analyses (optional)**
+   - Download enrichment libraries using `ov.utils.download_pathway_database()` and load them through `ov.utils.geneset_prepare`.
+   - Rank genes for GSEA with `rnk = dds.ranking2gsea()`.
+   - Instantiate `gsea_obj = ov.bulk.pyGSEA(rnk, pathway_dict)` and call `gsea_obj.enrichment()` to compute terms.
+   - Plot enrichment bubble charts via `gsea_obj.plot_enrichment(...)` and GSEA curves with `gsea_obj.plot_gsea(term_num=..., ...)`.
+8. **Troubleshooting**
+   - If PyDESeq2 raises errors about size factors, remind users to provide raw counts (not log-transformed data).
+   - `gene_id` mapping depends on species; direct them to download the correct genome pair when results look sparse.
+   - Large pathway libraries may require raising recursion limits or filtering to the top N terms before plotting.
+
+## Examples
+- "Run PyDESeq2 on treated vs control replicates and highlight the top enriched WikiPathways terms."
+- "Filter DEGs to genes with log2(BaseMean) > 1, auto-select fold-change cutoffs, and create volcano and boxplots."
+- "Generate the ranked gene list for GSEA and plot the enrichment curve for the top pathway."
+
+## References
+- Tutorial notebook: [`t_deseq2.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_deseq2.ipynb)
+- Sample featureCounts matrix: [`sample/counts.txt`](../../sample/counts.txt)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-stringdb-ppi/SKILL.md

@@ -0,0 +1,49 @@
+---
+name: string-protein-interaction-analysis-with-omicverse
+title: STRING protein interaction analysis with omicverse
+description: Help Claude query STRING for protein interactions, build PPI graphs with pyPPI, and render styled network figures for bulk gene lists.
+---
+
+# STRING protein interaction analysis with omicverse
+
+## Overview
+Invoke this skill when the user has a list of genes and wants to explore STRING protein–protein interactions via omicverse. The
+ workflow mirrors [`t_network.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_network.ipynb), covering species selection, S
+TRING API queries, and quick visualisation of the resulting network.
+
+## Instructions
+1. **Set up libraries**
+   - Import `omicverse as ov` and call `ov.utils.ov_plot_set()` (or `ov.plot_set()`) to match omicverse aesthetics.
+2. **Collect gene inputs**
+   - Accept a curated list of gene symbols (`gene_list = [...]`).
+   - Encourage the user to flag priority genes or categories so you can colour-code groups in the plot.
+3. **Assign metadata for plotting**
+   - Build dictionaries mapping genes to types and colours, e.g. `gene_type_dict = dict(zip(gene_list, ['Type1']*5 + ['Type2']*6
+))` and `gene_color_dict = {...}`.
+   - Remind users that consistent group labels improve legend readability.
+4. **Query STRING interactions**
+   - Call `ov.bulk.string_interaction(gene_list, species_id)` where `species_id` is the NCBI taxonomy ID (e.g. 4932 for yeast).
+   - Inspect the resulting DataFrame for combined scores and evidence channels to verify coverage.
+5. **Construct the network object**
+   - Initialise `ppi = ov.bulk.pyPPI(gene=gene_list, gene_type_dict=..., gene_color_dict=..., species=species_id)`.
+   - Run `ppi.interaction_analysis()` to fetch and cache STRING edges.
+6. **Visualise the network**
+   - Generate a default plot with `ppi.plot_network()` to reproduce the notebook figure.
+   - Mention that advanced styling (layout, node size, legends) can be tuned through `ov.utils.plot_network` keyword arguments if
+ the user requests adjustments.
+7. **Troubleshooting**
+   - Ensure gene symbols match the species—STRING expects case-sensitive identifiers; suggest mapping Ensembl IDs to symbols when
+ queries fail.
+   - If the API rate-limits, instruct the user to wait or provide a cached interaction table.
+   - For missing interactions, recommend enabling STRING's "add_nodes" option via `ppi.interaction_analysis(add_nodes=...)` to exp
+and the network.
+
+## Examples
+- "Retrieve STRING interactions for FAA4 and plot the network highlighting two gene classes."
+- "Download the STRING edge table for my Saccharomyces cerevisiae gene panel and colour nodes by module."
+- "Extend the network by adding the top five predicted partners before plotting."
+
+## References
+- Tutorial notebook: [`t_network.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_network.ipynb)
+- STRING background: [string-db.org](https://string-db.org/)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-to-single-deconvolution/SKILL.md

@@ -0,0 +1,50 @@
+---
+name: bulk-rna-seq-deconvolution-with-bulk2single
+title: Bulk RNA-seq deconvolution with Bulk2Single
+description: Turn bulk RNA-seq cohorts into synthetic single-cell datasets using omicverse's Bulk2Single workflow for cell fraction estimation, beta-VAE generation, and quality control comparisons against reference scRNA-seq.
+---
+
+# Bulk RNA-seq deconvolution with Bulk2Single
+
+## Overview
+Use this skill when a user wants to reconstruct single-cell profiles from bulk RNA-seq together with a matched reference scRNA-seq atlas. It follows [`t_bulk2single.ipynb`](../../omicverse_guide/docs/Tutorials-bulk2single/t_bulk2single.ipynb), which demonstrates how to harmonise PDAC bulk replicates, train the beta-VAE generator, and benchmark the output cells against dentate gyrus scRNA-seq.
+
+## Instructions
+1. **Load libraries and data**
+   - Import `omicverse as ov`, `scanpy as sc`, `scvelo as scv`, `anndata`, and `matplotlib.pyplot as plt`, then call `ov.plot_set()` to match omicverse styling.
+   - Read the bulk counts table with `ov.read(...)`/`ov.utils.read(...)` and harmonise gene identifiers via `ov.bulk.Matrix_ID_mapping(<df>, 'genesets/pair_GRCm39.tsv')`.
+   - Load the reference scRNA-seq AnnData (e.g., `scv.datasets.dentategyrus()`) and confirm the cluster labels (stored in `adata.obs['clusters']`).
+2. **Initialise the Bulk2Single model**
+   - Instantiate `ov.bulk2single.Bulk2Single(bulk_data=bulk_df, single_data=adata, celltype_key='clusters', bulk_group=['dg_d_1', 'dg_d_2', 'dg_d_3'], top_marker_num=200, ratio_num=1, gpu=0)`.
+   - Explain GPU selection (`gpu=-1` forces CPU) and how `bulk_group` names align with column IDs in the bulk matrix.
+3. **Estimate cell fractions**
+   - Call `model.predicted_fraction()` to run the integrated TAPE estimator, then plot stacked bar charts per sample to validate proportions.
+   - Encourage saving the fraction table for downstream reporting (`df.to_csv(...)`).
+4. **Preprocess for beta-VAE**
+   - Execute `model.bulk_preprocess_lazy()`, `model.single_preprocess_lazy()`, and `model.prepare_input()` to produce matched feature spaces.
+   - Clarify that the lazy preprocessing expects raw counts; skip if the user has already log-normalised data and instead provide aligned matrices manually.
+5. **Train or load the beta-VAE**
+   - Train with `model.train(batch_size=512, learning_rate=1e-4, hidden_size=256, epoch_num=3500, vae_save_dir='...', vae_save_name='dg_vae', generate_save_dir='...', generate_save_name='dg')`.
+   - Mention early stopping via `patience` and how to resume by reloading weights with `model.load('.../dg_vae.pth')`.
+   - Use `model.plot_loss()` to monitor convergence.
+6. **Generate and filter synthetic cells**
+   - Produce an AnnData using `model.generate()` and reduce noise through `model.filtered(generate_adata, leiden_size=25)`.
+   - Store the filtered AnnData (`.write_h5ad`) for reuse, noting it contains PCA embeddings in `obsm['X_pca']`.
+7. **Benchmark against the reference atlas**
+   - Plot cell-type compositions with `ov.bulk2single.bulk2single_plot_cellprop(...)` for both generated and reference data.
+   - Assess correlation using `ov.bulk2single.bulk2single_plot_correlation(single_data, generate_adata, celltype_key='clusters')`.
+   - Embed with `generate_adata.obsm['X_mde'] = ov.utils.mde(generate_adata.obsm['X_pca'])` and visualise via `ov.utils.embedding(..., color=['clusters'], palette=ov.utils.pyomic_palette())`.
+8. **Troubleshooting tips**
+   - If marker selection fails, increase `top_marker_num` or provide a curated marker list.
+   - Alignment errors typically stem from mismatched `bulk_group` names—double-check column IDs in the bulk matrix.
+   - Training on CPU can take several hours; advise switching `gpu` to an available CUDA device for speed.
+
+## Examples
+- "Estimate cell fractions for PDAC bulk replicates and generate synthetic scRNA-seq using Bulk2Single."
+- "Load a pre-trained Bulk2Single model, regenerate cells, and compare cluster proportions to the dentate gyrus atlas."
+- "Plot correlation heatmaps between generated cells and reference clusters after filtering noisy synthetic cells."
+
+## References
+- Tutorial notebook: [`t_bulk2single.ipynb`](../../omicverse_guide/docs/Tutorials-bulk2single/t_bulk2single.ipynb)
+- Example data and weights: [`omicverse_guide/docs/Tutorials-bulk2single/data/`](../../omicverse_guide/docs/Tutorials-bulk2single/data/)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-trajblend-interpolation/SKILL.md

@@ -0,0 +1,52 @@
+---
+name: bulktrajblend-trajectory-interpolation
+title: BulkTrajBlend trajectory interpolation
+description: Extend scRNA-seq developmental trajectories with BulkTrajBlend by generating intermediate cells from bulk RNA-seq, training beta-VAE and GNN models, and interpolating missing states.
+---
+
+# BulkTrajBlend trajectory interpolation
+
+## Overview
+Invoke this skill when users need to bridge gaps in single-cell developmental trajectories using matched bulk RNA-seq. It follows [`t_bulktrajblend.ipynb`](../../omicverse_guide/docs/Tutorials-bulk2single/t_bulktrajblend.ipynb), showcasing how BulkTrajBlend deconvolves PDAC bulk samples, identifies overlapping communities with a GNN, and interpolates "interrupted" cell states.
+
+## Instructions
+1. **Prepare libraries and inputs**
+   - Import `omicverse as ov`, `scanpy as sc`, `scvelo as scv`, and helper functions like `from omicverse.utils import mde`; run `ov.plot_set()`.
+   - Load the reference scRNA-seq AnnData (`scv.datasets.dentategyrus()`) and raw bulk counts with `ov.utils.read(...)` followed by `ov.bulk.Matrix_ID_mapping(...)` for gene ID harmonisation.
+2. **Configure BulkTrajBlend**
+   - Instantiate `ov.bulk2single.BulkTrajBlend(bulk_seq=bulk_df, single_seq=adata, bulk_group=['dg_d_1','dg_d_2','dg_d_3'], celltype_key='clusters')`.
+   - Explain that `bulk_group` names correspond to raw bulk columns and the method expects unscaled counts.
+3. **Set beta-VAE expectations**
+   - Call `bulktb.vae_configure(cell_target_num=100)` (or pass a dictionary) to define expected cell counts per cluster. Mention that omitting the argument triggers TAPE-based estimation.
+4. **Train or load the beta-VAE**
+   - Use `bulktb.vae_train(batch_size=512, learning_rate=1e-4, hidden_size=256, epoch_num=3500, vae_save_dir='...', vae_save_name='dg_btb_vae', generate_save_dir='...', generate_save_name='dg_btb')`.
+   - Highlight resuming with `bulktb.vae_load('.../dg_btb_vae.pth')` and the need to regenerate cells with consistent random seeds for reproducibility.
+5. **Generate synthetic cells**
+   - Produce filtered AnnData via `bulktb.vae_generate(leiden_size=25)` and inspect compositions with `ov.bulk2single.bulk2single_plot_cellprop(...)`.
+   - Save outputs to disk for reuse (`adata.write_h5ad`).
+6. **Configure and train the GNN**
+   - Call `bulktb.gnn_configure(max_epochs=2000, use_rep='X', neighbor_rep='X_pca', gpu=0, ...)` to set hyperparameters.
+   - Train using `bulktb.gnn_train()`; reload checkpoints with `bulktb.gnn_load('save_model/gnn.pth')`.
+   - Generate overlapping community assignments through `bulktb.gnn_generate()`.
+7. **Visualise community structure**
+   - Create MDE embeddings: `bulktb.nocd_obj.adata.obsm['X_mde'] = mde(bulktb.nocd_obj.adata.obsm['X_pca'])`.
+   - Plot clusters vs. discovered communities using `sc.pl.embedding(..., color=['clusters','nocd_n'], palette=ov.utils.pyomic_palette())` and filtered subsets excluding synthetic labels with hyphens.
+8. **Interpolate missing states**
+   - Run `bulktb.interpolation('OPC')` (replace with target lineage) to synthesise continuity, then preprocess the interpolated AnnData (HVG selection, scaling, PCA).
+   - Compute embeddings with `mde`, visualise with `ov.utils.embedding`, and compare to the original atlas.
+9. **Analyse trajectories**
+   - Initialise `ov.single.pyVIA` on both original and interpolated data to derive pseudotime, followed by `get_pseudotime`, `sc.pp.neighbors`, `ov.utils.cal_paga`, and `ov.utils.plot_paga` for topology validation.
+10. **Troubleshooting tips**
+    - If the VAE collapses (high reconstruction loss), lower `learning_rate` or reduce `hidden_size`.
+    - Ensure the same generated dataset is used before calling `gnn_train`; regenerating cells changes the graph and can break checkpoint loading.
+    - Sparse clusters may need adjusted `cell_target_num` thresholds or a smaller `leiden_size` filter to retain rare populations.
+
+## Examples
+- "Train BulkTrajBlend on PDAC cohorts, then interpolate missing OPC states in the trajectory."
+- "Load saved beta-VAE and GNN weights to regenerate overlapping communities and plot cluster vs. nocd labels."
+- "Run VIA on interpolated cells and compare PAGA graphs with the original scRNA-seq trajectory."
+
+## References
+- Tutorial notebook: [`t_bulktrajblend.ipynb`](../../omicverse_guide/docs/Tutorials-bulk2single/t_bulktrajblend.ipynb)
+- Example datasets and checkpoints: [`omicverse_guide/docs/Tutorials-bulk2single/data/`](../../omicverse_guide/docs/Tutorials-bulk2single/data/)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/bulk-wgcna-analysis/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: bulk-wgcna-analysis-with-omicverse
+title: Bulk WGCNA analysis with omicverse
+description: Assist Claude in running PyWGCNA through omicverse—preprocessing expression matrices, constructing co-expression modules, visualising eigengenes, and extracting hub genes.
+---
+
+# Bulk WGCNA analysis with omicverse
+
+## Overview
+Activate this skill for users who want to reproduce the WGCNA workflow from [`t_wgcna.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_wgcna.ipynb). It guides you through loading expression data, configuring PyWGCNA, constructing weighted gene co-expression networks, and inspecting modules of interest.
+
+## Instructions
+1. **Prepare the environment**
+   - Import `omicverse as ov`, `scanpy as sc`, `matplotlib.pyplot as plt`, and `pandas as pd`.
+   - Set plotting defaults via `ov.plot_set()`.
+2. **Load and filter expression data**
+   - Read expression matrices (e.g., from `expressionList.csv`).
+   - Calculate median absolute deviation with `from statsmodels import robust` and `gene_mad = data.apply(robust.mad)`.
+   - Keep the top variable genes (e.g., `data = data.T.loc[gene_mad.sort_values(ascending=False).index[:2000]]`).
+3. **Initialise PyWGCNA**
+   - Create `pyWGCNA_5xFAD = ov.bulk.pyWGCNA(name=..., species='mus musculus', geneExp=data.T, outputPath='', save=True)`.
+   - Confirm `pyWGCNA_5xFAD.geneExpr` looks correct before proceeding.
+4. **Preprocess the dataset**
+   - Run `pyWGCNA_5xFAD.preprocess()` to drop low-expression genes and problematic samples.
+5. **Construct the co-expression network**
+   - Evaluate soft-threshold power: `pyWGCNA_5xFAD.calculate_soft_threshold()`.
+   - Build adjacency and TOM matrices via `calculating_adjacency_matrix()` and `calculating_TOM_similarity_matrix()`.
+6. **Detect gene modules**
+   - Generate dendrograms and modules: `calculate_geneTree()`, `calculate_dynamicMods(kwargs_function={'cutreeHybrid': {...}})`.
+   - Derive module eigengenes with `calculate_gene_module(kwargs_function={'moduleEigengenes': {'softPower': 8}})`.
+   - Visualise adjacency/TOM heatmaps using `plot_matrix(save=False)` if needed.
+7. **Inspect specific modules**
+   - Extract genes from modules with `get_sub_module([...], mod_type='module_color')`.
+   - Build sub-networks using `get_sub_network(mod_list=[...], mod_type='module_color', correlation_threshold=0.2)` and plot them via `plot_sub_network(...)`.
+8. **Update sample metadata for downstream analyses**
+   - Load sample annotations `updateSampleInfo(path='.../sampleInfo.csv', sep=',')`.
+   - Assign colour maps for metadata categories with `setMetadataColor(...)`.
+9. **Analyse module–trait relationships**
+   - Run `analyseWGCNA()` to compute module–trait statistics.
+   - Plot module eigengene heatmaps and bar charts with `plotModuleEigenGene(module, metadata, show=True)` and `barplotModuleEigenGene(...)`.
+10. **Find hub genes**
+    - Identify top hubs per module using `top_n_hub_genes(moduleName='lightgreen', n=10)`.
+11. **Troubleshooting tips**
+    - Large datasets may require increasing `save=False` to avoid writing many intermediate files.
+    - If module detection fails, confirm enough genes remain after MAD filtering and adjust `deepSplit` or `softPower`.
+    - Ensure metadata categories have assigned colours before plotting eigengene heatmaps.
+
+## Examples
+- "Build a WGCNA network on the 5xFAD dataset, visualise modules, and extract hub genes from the lightgreen module."
+- "Load sample metadata, update colours for sex and genotype, and plot module eigengene heatmaps."
+- "Create a sub-network plot for the gold module using a correlation threshold of 0.2."
+
+## References
+- Tutorial notebook: [`t_wgcna.ipynb`](../../omicverse_guide/docs/Tutorials-bulk/t_wgcna.ipynb)
+- Tutorial dataset: [`data/5xFAD_paper/`](../../omicverse_guide/docs/Tutorials-bulk/data/5xFAD_paper/)
+- Quick copy/paste commands: [`reference.md`](reference.md)

package/data/skills/cancer-metabolism-agent/SKILL.md

@@ -0,0 +1,180 @@
+<!--
+# COPYRIGHT NOTICE
+# This file is part of the "Universal Biomedical Skills" project.
+# Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>
+# All Rights Reserved.
+#
+# This code is proprietary and confidential.
+# Unauthorized copying of this file, via any medium is strictly prohibited.
+#
+# Provenance: Authenticated by MD BABU MIA
+
+-->
+
+---
+name: 'cancer-metabolism-agent'
+description: 'AI-powered analysis of cancer metabolic reprogramming including Warburg effect, glutamine addiction, lipid metabolism, and metabolic vulnerabilities for therapeutic targeting.'
+measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes.
+allowed-tools:
+  - read_file
+  - run_shell_command
+---
+
+
+# Cancer Metabolism Agent
+
+The **Cancer Metabolism Agent** analyzes tumor metabolic reprogramming to identify vulnerabilities for therapeutic targeting. It integrates metabolomics, transcriptomics, and flux analysis to characterize Warburg effect, glutamine addiction, lipid synthesis, and other cancer-specific metabolic alterations.
+
+## When to Use This Skill
+
+* When analyzing tumor metabolomic profiles to identify metabolic phenotypes.
+* To identify metabolic vulnerabilities as therapeutic targets.
+* For predicting response to metabolism-targeting drugs (metformin, 2-DG, CB-839).
+* When integrating metabolomics with transcriptomics for pathway analysis.
+* To analyze tumor-microenvironment metabolic competition.
+
+## Core Capabilities
+
+1. **Metabolic Phenotyping**: Classify tumors by dominant metabolic programs (glycolytic, oxidative, lipogenic).
+
+2. **Warburg Effect Quantification**: Measure aerobic glycolysis and lactate production signatures.
+
+3. **Glutamine Dependency Analysis**: Identify glutamine-addicted tumors vulnerable to GLS inhibitors.
+
+4. **Lipid Metabolism Profiling**: Analyze de novo lipogenesis and fatty acid oxidation.
+
+5. **Metabolic Flux Analysis**: Integrate 13C tracer data for pathway flux quantification.
+
+6. **Drug Sensitivity Prediction**: Predict response to metabolism-targeting therapeutics.
+
+## Key Metabolic Pathways in Cancer
+
+| Pathway | Key Enzymes | Cancer Relevance | Therapeutic Targets |
+|---------|-------------|------------------|---------------------|
+| Glycolysis | HK2, PKM2, LDHA | Warburg effect | 2-DG, lonidamine |
+| Glutaminolysis | GLS1, GDH | Nitrogen/carbon source | CB-839, BPTES |
+| Fatty acid synthesis | FASN, ACC, ACLY | Membrane biogenesis | TVB-2640, ND-646 |
+| Oxidative phosphorylation | Complex I-V | OXPHOS tumors | Metformin, IACS-010759 |
+| One-carbon metabolism | SHMT, MTHFD | Nucleotide synthesis | Methotrexate |
+| Serine synthesis | PHGDH, PSAT1 | Amino acid auxotrophy | NCT-503 |
+
+## Workflow
+
+1. **Input**: Metabolomics data (LC-MS, GC-MS), RNA-seq expression, clinical annotations.
+
+2. **Normalization**: Process metabolomics data with appropriate normalization.
+
+3. **Pathway Scoring**: Calculate metabolic pathway activity scores.
+
+4. **Phenotype Classification**: Assign metabolic phenotype clusters.
+
+5. **Vulnerability Identification**: Identify metabolic dependencies.
+
+6. **Drug Matching**: Predict sensitivity to metabolism-targeting agents.
+
+7. **Output**: Metabolic phenotype, pathway activities, therapeutic recommendations.
+
+## Example Usage
+
+**User**: "Analyze this tumor's metabolic profile and identify targetable metabolic vulnerabilities."
+
+**Agent Action**:
+```bash
+python3 Skills/Oncology/Cancer_Metabolism_Agent/metabolism_analyzer.py \
+    --metabolomics tumor_lcms.csv \
+    --rnaseq tumor_expression.tsv \
+    --tumor_type NSCLC \
+    --normalize mtic \
+    --pathway_analysis true \
+    --drug_prediction true \
+    --output metabolism_report/
+```
+
+## Metabolic Phenotype Classification
+
+**Glycolytic (Warburg)**:
+- High HK2, PKM2, LDHA expression
+- Elevated lactate/pyruvate ratio
+- Low mitochondrial gene expression
+- Sensitive to glycolysis inhibitors
+
+**Oxidative (OXPHOS-dependent)**:
+- High ETC complex expression
+- Active TCA cycle
+- PGC1α driven
+- Sensitive to metformin, IACS-010759
+
+**Lipogenic**:
+- High FASN, ACC, SREBP1/2
+- Active de novo lipogenesis
+- Common in prostate, breast cancer
+- Sensitive to FASN inhibitors
+
+**Glutamine-addicted**:
+- High GLS1, MYC-driven
+- Glutamine-dependent anaplerosis
+- Common in KRAS-mutant cancers
+- Sensitive to CB-839
+
+## AI/ML Models
+
+**Metabolic Phenotype Classifier**:
+- Random forest on metabolite ratios
+- 85% accuracy on validation cohorts
+- Integrates with molecular subtypes
+
+**Flux Balance Analysis**:
+- Genome-scale metabolic models (Recon3D)
+- Constraint-based optimization
+- Predicts essential metabolic genes
+
+**Drug Response Prediction**:
+- GDSC/CCLE metabolic drug data
+- Multi-omic feature integration
+- AUC 0.75-0.85 for metabolic drugs
+
+## Metabolomics Data Processing
+
+| Step | Method | Purpose |
+|------|--------|---------|
+| Peak detection | XCMS, MZmine | Identify metabolites |
+| Annotation | HMDB, KEGG | Assign identities |
+| Normalization | MTIC, median | Remove batch effects |
+| Imputation | KNN, RF | Handle missing values |
+| Enrichment | MSEA, Mummichog | Pathway analysis |
+
+## TME Metabolic Competition
+
+The agent analyzes tumor-immune metabolic crosstalk:
+- Glucose competition (T-cell activation)
+- Lactate immunosuppression
+- Arginine depletion by MDSCs
+- Tryptophan-IDO axis
+- Adenosine immunosuppression
+
+## Prerequisites
+
+* Python 3.10+
+* COBRApy for flux balance
+* MetaboAnalyst interface
+* Pathway databases (KEGG, Reactome)
+
+## Related Skills
+
+* Metabolomics_Agent - For general metabolomics
+* Multi_Omics_Integration - For omic integration
+* Drug_Repurposing - For therapeutic matching
+
+## Clinical Applications
+
+1. **Treatment Selection**: Match metabolic phenotype to drugs
+2. **Combination Therapy**: Identify synergistic metabolic targets
+3. **Resistance Mechanisms**: Metabolic adaptation under therapy
+4. **Diet Interventions**: Ketogenic diet in glycolytic tumors
+
+## Author
+
+AI Group - Biomedical AI Platform
+
+
+<!-- AUTHOR_SIGNATURE: 9a7f3c2e-MD-BABU-MIA-2026-MSSM-SECURE -->

package/data/skills/care-coordination/SKILL.md

@@ -0,0 +1,35 @@
+<!--
+# COPYRIGHT NOTICE
+# This file is part of the "Universal Biomedical Skills" project.
+# Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>
+# All Rights Reserved.
+#
+# This code is proprietary and confidential.
+# Unauthorized copying of this file, via any medium is strictly prohibited.
+#
+# Provenance: Authenticated by MD BABU MIA
+
+-->
+
+---
+name: 'care-coordination'
+description: 'Care Coordination agent for healthcare workflows.'
+measurable_outcome: Execute skill workflow successfully with valid output within 15 minutes.
+allowed-tools:
+  - read_file
+  - run_shell_command
+---
+
+
+# Care Coordination
+
+This skill implements the Care Coordination workflow using Anthropic's Claude.
+
+## Usage
+
+```bash
+python3 Skills/Anthropic_Health_Stack/Care_Coordination/coworker.py
+```
+
+
+<!-- AUTHOR_SIGNATURE: 9a7f3c2e-MD-BABU-MIA-2026-MSSM-SECURE -->