exploratory-data-analysis

Exploratory Data Analysis (探索性数据分析)

Skill Overview

An exploratory data analysis tool that automatically analyzes 200+ scientific data file formats, auto-detects file types, and generates detailed reports that include quality assessments and recommendations for downstream analyses.

Use Cases

1. Experimental Data Quality Checks

Quickly assess data completeness, consistency, and quality metrics before starting formal analyses to discover potential issues such as missing values, outliers, or format errors, avoiding wasted compute resources discovered late in the analysis process.

2. Analysis of Files with Uncertain Formats

When encountering unfamiliar scientific data formats or files downloaded from public databases with unknown formats, the skill will automatically identify the file type, extract structural information, and advise how to handle such data.

3. Cross-Disciplinary Data Integration

Handle datasets from different experimental platforms and disciplines (for example, sequencing data, microscopy images, and chemical structure files simultaneously), providing a unified view of each dataset’s characteristics to prepare for integrative analyses.

Core Features

Automatic Format Detection and Classification

Supports 200+ scientific data formats across chemistry and molecular formats (PDB, MOL, SDF), bioinformatics formats (FASTA, FASTQ, VCF, BAM), microscopy imaging formats (ND2, CZI, OME-TIFF), spectroscopic formats (mzML, DX, JDX), proteomics/metabolomics formats, and general scientific formats (CSV, HDF5, NetCDF).

Format-Specific Analysis

Performs targeted in-depth analyses based on detected file types: computes GC content and quality score distributions for sequence data, validates bond lengths and angles for structure files, extracts dimension and channel information for imaging data, computes statistical summaries and correlations for tabular data—delivering truly "format-aware" analysis.

Detailed Report Generation

Automatically generates a complete Markdown report including file metadata, structural information, statistical summaries, quality assessments, visualization suggestions, and recommendations for downstream analysis methods. Reports are ready for use in experimental records, methods documentation, or team collaboration.

Frequently Asked Questions

Which scientific data file formats does this skill support?

Supports 200+ formats across six main categories: Chemistry & molecular (60+ formats, e.g., PDB, MOL, CIF, XYZ), Bioinformatics & genomics (50+ formats, e.g., FASTA/Q, BAM, VCF, GFF), Microscopy & imaging (45+ formats, e.g., ND2, CZI, OME-TIFF), Spectroscopy & analytical chemistry (35+ formats, e.g., mzML, DX, SPC), Proteomics & metabolomics (30+ formats, e.g., mzML, mzTab, mzIdentML), and General scientific data (30+ formats, e.g., CSV, HDF5, Zarr).

What does the generated analysis report include?

The report contains eight core sections: basic file information (size, location, timestamps), format identification and description, data structure dimensions, statistical summaries, data quality assessment (missing values, outliers, consistency checks), visualization recommendations, applicable downstream analysis methods, and recommended Python libraries. All content is presented in structured Markdown for easy reading and version control.

How long does it take to analyze large data files?

For most files, analysis completes within seconds to minutes. For exceptionally large files (e.g., sequencing data tens of GB or large microscopy images), the skill uses intelligent sampling strategies to analyze representative samples and provide estimates, avoiding out-of-memory issues or long waits. The report will clearly state whether the analysis was performed on the full dataset or on sampled data.

Can this skill replace specialized analysis software?

Not entirely. This skill focuses on exploratory data analysis (EDA)—understanding what the data looks like, its quality, and what analyses it is suitable for—rather than performing deep, domain-specific analyses (e.g., differential expression, molecular docking, image segmentation). Its value lies in rapid assessment and planning, helping you choose appropriate downstream specialized tools.