3. Data, Science & AI

Transcribes audio/video to text with timestamps, speaker labels, and chapters from YouTube or local files, producing structured markdown output.

Creates and maintains structured AI instruction files (CLAUDE.md, .cursorrules) for consistent AI tool configuration and system prompt management.

Enables writing dataframe-agnostic Python code that works across libraries like pandas and polars with correct type annotations.

Runs AI model evaluations on SWE-bench and other datasets using the mcpbr benchmarking tool.

Provides a quick-start command to run SWE-bench Lite, an AI evaluation benchmark for assessing code generation and software engineering capabilities.

Clusters TCR/BCR clones by CDR3 sequences using Faiss-based algorithms, adding a cluster column for clonotype analysis in immunology.

Executes fast Gene Set Enrichment Analysis (GSEA) on single-cell RNA-seq data via fgsea R package, identifying enriched pathways and producing publication-ready visualizations.

Identifies differentially expressed marker genes across unsupervised cell clusters to determine broad cell types in immune cell populations.

Initial coarse clustering of all cell types (T, B, non-T/B) to identify broad populations before T/B cell selection in single-cell RNA-seq analysis.

Maps single-cell RNA-seq query datasets to reference atlases via Seurat's workflow, enabling label transfer and integration without modifying query data.

Performs pseudo-bulk differential gene expression analysis using DESeq2 or edgeR, aggregating single-cell data to identify differentially expressed genes while accounting for experimental design complexities.

Analyzes physicochemical properties of CDR3 amino acid sequences using regression to compare T-cell receptor repertoires between cell groups.

Annotates single-cell clusters with biological cell types using AI-driven methods including CellTypist and ScType for data interpretation.

Generates clonality statistics and diversity visualizations for TCR/BCR repertoire analysis, including Shannon, Simpson, and Gini metrics.

Separates T/B cells from mixed populations using VDJ data, marker genes, or k-means clustering for immunological analysis.

Integrates scTCR/BCR repertoire data with scRNA-seq expression data into Seurat objects for clonotype-aware gene expression analysis.

Performs metabolic pathway enrichment analysis using GSEA-based methods with fgsea, generating visualizations for biological interpretation across cell groups.

Performs unsupervised clustering on single-cell RNA-seq data using Seurat, including UMAP visualization and Louvain/Leiden algorithms for cell cluster identification.

Loads and processes single-cell TCR-seq and scBCR-seq data from multiple formats into scRepertoire-compatible objects for scRNA-seq integration.

Visualizes cell-cell communication inference results with network diagrams, heatmaps, and interaction plots for interpreting ligand-receptor interactions between cell types.

Generates cluster statistics and visualizations for Seurat single-cell RNA-seq data, including dimension reduction and gene expression plots.

Analyzes metabolic pathway heterogeneity in cell populations using NES, PCA, and GSEA to quantify diversity and identify subpopulation-specific metabolic states.

Identifies top-expressed genes per cell cluster and conducts pathway enrichment analysis to visualize gene expression patterns and biological functions across cell populations.

Comprehensive metabolic pathway analysis pipeline for scRNA-seq data, including expression imputation, feature selection, and heterogeneity analysis.