Skill UI

This advanced research technique is designed for deep dives into football topics. Instead of relying solely on the original query, it mandates tracking every discovered entity—such as player names, club names, or coaches—from search results. By treating these entities as new leads, the user must formulate follow-up searches using the discovered names and remaining constraints, ensuring nothing relevant is overlooked and building a comprehensive knowledge chain.

AnnData is a core Python framework designed for managing and analyzing annotated matrices, essential in single-cell genomics and high-dimensional data science. It provides a standardized structure (X, obs, var) to store experimental measurements alongside rich metadata. Users can read/write various formats (h5ad, zarr), perform subsetting, concatenation, and complex data manipulation, making it indispensable for modern bioinformatics workflows.

Provides programmatic and standardized access to massive, versioned datasets of single-cell and spatial transcriptomics from CZ CELLxGENE. This skill allows users to query millions of cells, analyze gene expression, and access pre-calculated embeddings across various tissues, diseases, and species without downloading entire datasets. Essential for large-scale bioinformatics analysis, machine learning model training, and comprehensive cross-dataset comparison.

Dask is a powerful Python library designed for scalable, parallel, and distributed computing. It enables workflows for datasets that exceed available RAM or require processing across multiple machines. Users can scale familiar pandas and NumPy operations (DataFrames, Arrays) and process unstructured data (Bags) efficiently, making it ideal for large-scale ETL, analytics, and distributed machine learning tasks.

Datamol is a high-level, Pythonic abstraction layer built over RDKit, designed for computational chemistry and drug discovery. It simplifies complex molecular operations, offering sensible defaults for tasks like SMILES standardization, descriptor calculation, fingerprint generation (ECFP4), 3D conformer analysis, and molecular clustering. It supports efficient parallel processing and cloud storage I/O, making it ideal for handling large-scale chemical datasets.

FlowIO is a specialized, low-level library designed for reading, inspecting, and writing Flow Cytometry Standard (FCS) files (versions 2.0, 3.0, and 3.1). It allows users to extract event data as NumPy arrays, inspect detailed metadata (HEADER, TEXT, ANALYSIS), and handle multi-dataset structures. It is crucial for data preparation, format conversion, and foundational metadata management in bioinformatics pipelines, but does not perform complex analysis like gating or compensation.

Geniml provides a robust framework for running machine learning and statistical workflows over genomic interval sets. It treats coordinates, assemblies, and model artifacts as explicit contracts, enabling users to validate data structures (like BED files) and plan complex runs (such as Region2Vec or scEmbed) locally. The tool emphasizes reproducibility, adherence to biological data standards, and running without unnecessary network calls.

A comprehensive guide for advanced geospatial data processing using GeoPandas. This skill targets professional workflows involving planar vector data, emphasizing rigorous data provenance, coordinate reference system (CRS) management, and complex spatial operations. It details necessary safety contracts, correctness gates, and best practices for handling sensitive geographic information in reproducible Python environments.

The HypoGeniC skill facilitates the iterative proposal and scoring of textual hypotheses from labeled datasets using LLMs. It provides tools for generating candidate hypothesis banks and task-prediction statistics. Crucially, it mandates a rigorous, local review workflow, emphasizing dataset auditing (checksums, schemas), configuration validation, and explicitly separating candidate results from definitive scientific conclusions.

LaminDB is an open-source, lineage-native lakehouse designed for managing complex biological datasets and computational workflows. It makes diverse scientific data (like scRNA-seq, AnnData) queryable, traceable, and compliant with FAIR principles. Core capabilities include robust data lineage tracking, schema validation, ontology-backed annotation (Bionty), and unified querying across multiple cloud/local storage systems. It is essential for bioinformatics research and building reproducible data infrastructure.

This skill provides a comprehensive framework for building, registering, debugging, and operating complex bioinformatics workflows on the Latch platform. It supports multiple industry-standard pipelines (Nextflow, Snakemake) and utilizes the Python SDK, CLI, and Latch Data/Registry. Use this when developing, deploying, or managing robust, data-intensive computational pipelines in a controlled environment.

A comprehensive guide for designing, reviewing, and migrating numerical workflows in MATLAB or GNU Octave. It emphasizes building reproducible projects, adhering to strict safety boundaries, and ensuring code semantics are explicit for robust data processing and analysis.