Skip to content

CLAUDE.md

Latest commit

 

History

History
109 lines (82 loc) · 3.29 KB

CLAUDE.md

File metadata and controls

109 lines (82 loc) · 3.29 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

TagDust2 is a bioinformatics tool for processing next-generation sequencing (NGS) data. It extracts and labels sequences containing adapter, linker, barcode, and fingerprint sequences using Hidden Markov Models (HMMs). The tool can demultiplex reads, automatically detect library preparation methods, and filter out contaminants.

Build System

This project uses GNU Autotools for configuration and building:

Essential Build Commands

# Initial setup (run once after cloning)
./autogen.sh
./configure

# Build the project
make

# Run tests
make check

# Install system-wide
make install

Development Build Options

# Enable debugging
./configure --enable-debugging

# Enable valgrind tests
./configure --enable-valgrind-tests

# Build with specific compiler flags
./configure CFLAGS="-O2 -Wall -std=gnu99"

Testing

  • Main test suite: make check
  • Individual test programs are built in src/: tagdustiotest, tagdust_rtest, simreads_rtest, evalres_rtest
  • Development tests in dev/: sanity_test.sh, casava_test.sh, bar_read_test.sh

Architecture

Core Components

  • main.c: Main tagdust program entry point
  • barcode_hmm.c/h: HMM construction, training, and sequence matching
  • interface.c/h: Command-line argument parsing and program initialization
  • io.c/h: File I/O operations for FASTQ/FASTA formats
  • nuc_code.c/h: Nucleotide encoding and sequence operations
  • misc.c/h: Utility functions and data structures
  • kslib.c/h: Sequence parsing and utility library

Executables Built

  1. tagdust: Main demultiplexing tool
  2. simreads: Read simulation utility
  3. evalres: Results evaluation tool
  4. merge: Sequence merging utility
  5. rename_qiime: QIIME format renaming tool

Key Data Structures

  • HMM structures for sequence architecture modeling
  • Sequence data structures with quality scores
  • Barcode matching and scoring systems

Configuration Files

Architecture Files

Architecture files define the expected structure of reads:

  • Located in dev/ and casava_demo/ directories
  • Examples: casava_arch.txt, various EDITTAG_* files
  • Format: Specify barcodes, read segments, and expected patterns

Test Data

  • dev/: Contains test FASTQ files and expected output files
  • benchmark/: Performance testing scripts and data
  • Gold standard files ending in _gold.txt for validation

Development Workflow

Adding New Features

  1. Modify relevant source files in src/
  2. Update src/Makefile.am if adding new source files
  3. Run make to build
  4. Run make check to ensure tests pass
  5. Add new tests in dev/ if needed

Debugging

  • Configure with --enable-debugging for debug builds
  • Use --enable-debugging=1|2|3 for different verbosity levels
  • Debug macros available via debug_print() in tagdust2.h

Memory Management

  • Custom memory allocation macros in malloc_macro.h
  • SIMD-aligned memory allocation support
  • Memory debugging available with debug builds

Paper Reproducibility

To reproduce the benchmarks from the TagDust2 paper:

cd reproducibility/scripts
make -f run.mk benchmark

This requires 64-bit system and may take considerable time.