Omics Analysis Skills Index: анализ данных, скилл

--- id: omics_skills_index name: Omics Analysis Skills Index description: | Skills for single-cell and spatial omics data analysis. Best practices, code snippets, and workflows for the scverse ecosystem. --- # Agent Skills for Omics Data Analysis Here are best practices and workflows for analyzing single-cell and spatial omics data. When performing specific analysis tasks, load the relevant skill files to guide your approach. ## Available Skills ### Single-Cell to Spatial Mapping If you have both single-cell and spatial data for the same/similar sample, you can map single-cell data to spatial data to impute unobserved genes and enhance spatial resolution. **Skill file**: [single_cell_spatial_mapping.md](./single_cell_spatial_mapping.md) **When to use**: - You have paired scRNA-seq and spatial transcriptomics data - You want to impute genes not measured in the spatial modality - You want to transfer cell type annotations to spatial coordinates --- ### 3D Spatial Data Visualization For visualizing 3D spatial transcriptomics data with interactive plots and animations. **Skill file**: [visualize_3d_spatial.md](./visualize_3d_spatial.md) **When to use**: - Your spatial data has 3D coordinates - You want to visualize gene expression or cell types in 3D - You want to create rotating GIF animations --- ### Quality Control Workflow Standard quality control workflow for single-cell data. **Skill file**: [quality_control.md](./quality_control.md) **When to use**: - Starting analysis of new single-cell dataset - Need to filter low-quality cells - Assessing data quality metrics --- ### Cell Type Annotation Approaches for annotating cell types in single-cell data. **Skill file**: [cell_type_annotation.md](./cell_type_annotation.md) **When to use**: - After clustering, need to assign cell type labels - Using marker genes for annotation - Using reference-based methods --- ### Single-Cell Foundation Models (SCFM) Workflow and model reference for embedding/integration with foundation models (scGPT, Geneformer, UCE). **Skill files**: - [scfm-workflow.md](file:///scfm-workflow.md) - [scfm-models.md](file:///scfm-models.md) **When to use**: - You want FM embeddings (e.g., `obsm["X_uce"]`, `obsm["X_scGPT"]`, `obsm["X_geneformer"]`) - You need model selection based on gene ID scheme and species - You want a validation-first workflow before heavy inference --- ### Trajectory Inference Pseudotime analysis and trajectory inference for cell differentiation, neurogenesis, and lineage tracing studies. **Skill file**: [trajectory_inference.md](./trajectory_inference.md) **When to use**: - Studying cell differentiation paths (e.g., stem cell → mature cell) - Neurogenesis analysis (neural progenitors → neurons) - Comparing developmental trajectories between conditions - RNA velocity analysis for directional dynamics --- ### Parallel Computing & Performance Strategies for accelerating single-cell analysis using multi-core CPU, GPU acceleration, and memory optimization. **Skill file**: [parallel_computing.md](./parallel_computing.md) **When to use**: - Analysis is running slowly on single core - Dataset has >50k cells and operations are timing out - GPU is available and you want 10-100x speedup - Need to parallelize custom analysis loops --- ## Upstream Processing Technology-specific pipelines for processing raw sequencing data into analysis-ready count matrices with spatial coordinates. These cover the steps that precede standard single-cell analysis (QC, normalization, clustering, etc.). **Skill index**: [upstream_processing/SKILL.md](./upstream_processing/SKILL.md) **Technologies covered**: - **nf-core Pipelines**: 143+ curated Nextflow pipelines for scRNA-seq, spatial transcriptomics, bulk RNA-seq, ATAC-seq, ChIP-seq, CUT&Run, methylation, and variant calling (WGS/WES) - **OpenST**: Open-source spatial transcriptomics at sub-cellular resolution — flow cell barcode preprocessing, spacemake alignment, image registration, Cellpose segmentation, 3D reconstruction, and downstream analysis **When to use**: - Processing raw BCL/FASTQ files with nf-core community pipelines - Running technology-specific alignment and preprocessing pipelines - Spatial coordinate registration and cell segmentation - Variant calling from WGS/WES/targeted sequencing - 3D reconstruction from serial tissue sections --- ## Database Access Tools for querying genomic databases, downloading sequencing data from public repositories, and accessing large-scale single-cell datasets programmatically. **Skill index**: [database_access/SKILL.md](./database_access/SKILL.md) **Tools covered**: - **gget**: Python package with 23 modules for querying Ensembl, NCBI, UniProt, ARCHS4, Enrichr, COSMIC, OpenTargets, CellxGene, cBioPortal, PDB, and Bgee - **iSeq**: Bash CLI for downloading sequencing data from GSA, SRA, ENA, DDBJ, and GEO databases with parallel downloads and Aspera support - **CZ CELLxGENE Census**: Cloud-based Python API for accessing 217M+ single-cell RNA-seq observations with flexible metadata queries and pre-computed embeddings **When to use**: - Querying gene/protein information from public databases - Downloading raw sequencing data (FASTQ/SRA) from public repositories - Accessing curated single-cell RNA-seq datasets by tissue, cell type, disease - Performing enrichment analysis or cancer mutation queries - Fetching reference genomes, annotations, and sequences --- ## Supplementary Reference: SC Best Practices For more comprehensive guidance on single-cell and spatial omics analysis, refer to the **SC Best Practices** skill collection, derived from the authoritative [Single-cell Best Practices](https://www.sc-best-practices.org) book. It covers the full analysis pipeline from preprocessing to reproducibility, including detailed workflows, method comparisons, and code examples for the scverse ecosystem. **Skill index**: [sc_best_practices/SKILL.md](./sc_best_practices/SKILL.md) **Topics covered**: - Introduction & raw data processing frameworks - Preprocessing (QC, normalization, HVG, dimensionality reduction) - Clustering, annotation & dataset integration - Trajectory analysis (pseudotime, RNA velocity, lineage tracing) - Differential expression & condition analysis - Gene regulatory networks & cell-cell communication - Bulk deconvolution, chromatin accessibility (scATAC-seq) - Spatial omics (neighborhood analysis, deconvolution, imputation) - Surface protein (CITE-seq), immune repertoire (TCR/BCR) - Multimodal integration & reproducibility When the skills above provide task-specific workflows, these supplementary references offer broader context, alternative methods, and detailed best practices to complement your analysis. --- ## Using Skills 1. **Before analysis**: Scan this index for relevant skills 2. **Load skill file**: Read the full skill document for detailed guidance 3. **Follow best practices**: Use the code snippets and workflows provided 4. **Adapt as needed**: Skills are templates; adjust for your specific data