Skip to content

PROFILE_RECOMMEND.md

Latest commit

 

History

History
90 lines (65 loc) · 2.02 KB

PROFILE_RECOMMEND.md

File metadata and controls

90 lines (65 loc) · 2.02 KB

RAPTOR Profile & Recommend Guide

Get intelligent pipeline recommendations in 30 seconds!

Quick Start

# Basic profiling
raptor profile --counts your_data.csv

# With metadata  
raptor profile --counts data.csv --metadata samples.csv

# Save HTML report
raptor profile --counts data.csv --output report.html

How It Works

  1. Analyzes your data: BCV, depth, zeros, library sizes
  2. Scores all 8 pipelines: Accuracy, speed, memory, compatibility
  3. Recommends top 3: With detailed reasoning
  4. Generates report: Interactive HTML with visualizations

Understanding Output

#1 RECOMMENDED: Pipeline 3 (Salmon-edgeR)
   Score: 0.88 | Confidence: High
   
   Why?
   ✓ Excellent balance for your data
   ✓ Handles medium BCV well (yours: 0.42)
   ✓ Fast runtime (~1hr for 12 samples)
   ✓ Low memory (8GB)

Key Metrics Explained

BCV (Biological Coefficient of Variation)

  • Low (<0.2): Cell lines, controlled
  • Medium (0.2-0.6): Typical studies
  • High (>0.6): Clinical, complex

Sequencing Depth

  • Low (<10M): May miss genes
  • Medium (10-25M): Adequate
  • High (>25M): Excellent

Advanced Usage

# Prioritize accuracy
raptor profile --counts data.csv \
  --weight-accuracy 0.7 --weight-speed 0.1

# Resource constraints
raptor profile --counts data.csv \
  --max-memory 16G --max-runtime 4h

# Exclude pipelines
raptor profile --counts data.csv \
  --exclude-pipelines 7,8

Python API

from raptor import RNAseqDataProfiler, PipelineRecommender

# Profile
profiler = RNAseqDataProfiler(counts, metadata)
profile = profiler.profile()

# Recommend
recommender = PipelineRecommender()
recommendations = recommender.recommend(profile)

Examples

Standard DE study (12 samples, 20M reads):
→ Recommend: Pipeline 3 (Salmon-edgeR)

Large cohort (100 samples):
→ Recommend: Pipeline 4 (Kallisto-Sleuth)

Small pilot (4 samples, need accuracy):
→ Recommend: Pipeline 1 (STAR-RSEM-DESeq2)

See full examples and interpretation guide in complete documentation.