README: Spliced Alignment Benchmarking Resource

Aligning transcripts to their genomic source locations is a surprisingly difficult problem. This project seeks to answer the following questions:

• Do aligners fail some simple sanity checks?
• Which aligner is the most accurate?
• Which aligner is the most efficient?
• What kinds of sequence and gene features create the most problems?
• What improvements can be made in spliced alignment?

Quickstart

1. Install conda (e.g. Miniforge3)
2. Clone this repo
3. Create conda environment
4. Run the demos in the bakeoff usage statement

See the TUTORIAL.md for a step-by-step walkthrough.

Manifest

• README.md this document
• TUTORIAL.md a quick walk-through to check that things work
• INFO.md some behind-the-scenes information
• NOTES.md random stuff the devs are thinking about or working on
• bakeoff top-level program for assessing aligners
• env/ directory of conda environments for different platforms
• data/ directory with some sample files (1% of favorite genomes)
• src/ directory with programs that run various parts of the analyses
• 2025/ directory with specifics for the 2025 study

estgenome

Testing

conda activate sabr-linux-x86
python3 src/read-simulator.py data/ce01.fa.gz data/ce01.ftx.gz --seed 1 --samplegene 0.05 --samplereads 0.01 > rtest.fa
python3 src/est-genome.py data/ce01.fa.gz rtest.fa --verbose

Runnning

• Remove `--verbose`` flag
• Add --threads n for however many CPUs you are using
• Remove temp directory (it doesn't auto-delete for debugging reasons)