Sovereign 8s.is mirror of standardgalactic/sparse_features
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2025-12-18 15:41:44 -08:00
data add nnz u64 support 2025-12-18 14:35:13 -08:00
src add nnz u64 support 2025-12-18 15:41:44 -08:00
Cargo.lock add nnz u64 support 2025-12-18 14:34:46 -08:00
Cargo.toml bug fix 2025-11-25 21:27:22 -08:00
README.md bug fix 2025-11-25 21:27:22 -08:00

Simulate sparse features based on feature name, sparsity for any number of samples

Usage

Simulate sparse OTU/feature tables from a Newick tree and write BIOM (HDF5 CSR+CSC)

Usage: sparse_features [OPTIONS] --tree <tree> --samples <samples>

Options:
  -t, --tree <tree>
          Input Newick tree file (taxa from leaf names)
  -n, --samples <samples>
          Number of samples (columns)
      --mean-sparsity <mean_sparsity>
          Average per-sample nonzero fraction (e.g., 0.02 = 2%) [default: 0.02]
      --std-sparsity <std_sparsity>
          Stddev of per-sample sparsity (Normal, then clipped) [default: 0.005]
      --min-sparsity <min_sparsity>
          Lower clip for per-sample sparsity [default: 0.0]
      --max-sparsity <max_sparsity>
          Upper clip for per-sample sparsity [default: 0.5]
      --min-count <min_count>
          Minimum count for a nonzero entry [default: 1]
      --max-count <max_count>
          Maximum count for a nonzero entry [default: 100]
      --chunk-size <chunk_size>
          Number of taxa to process per chunk (parallelized) [default: 2048]
      --max-nz-per-row <max_nz_per_row>
          Optional hard cap of nonzeros per taxon (prevents giant rows)
      --seed <seed>
          Base RNG seed (u64) for reproducibility [default: 42]
  -o, --output <output>
          Output BIOM (HDF5) path [default: simulated.biom]
      --sample-prefix <sample_prefix>
          Sample ID prefix (IDs will be prefix1..prefixN) [default: sample]
  -h, --help
          Print help
  -V, --version

Install

git clone https://github.com/jianshu93/sparse_features
cd sparse_features
cargo build --release
./target/release/sparse_features -h

Generating feature tables

 sparse_features -t ../../data/ASVs_aligned.tre -o ./ASVs_otu.biom --samples 1000 --min-sparsity 
0.0005 --max-count 3000 --chunk-size 1000 --mean-sparsity 0.03

Ideally --min-sparsity should be >= ln(n_cols) (sample) / n_rows (taxa) to avoid empty samples.