Skip to content

input-formats.md

Latest commit

 

History

History
84 lines (58 loc) · 2.54 KB

input-formats.md

File metadata and controls

84 lines (58 loc) · 2.54 KB

Input Format Reference

Training FASTA (train)

Standard FASTA format. The first whitespace-delimited token in each header is used as the class label.

>L4 sample_0001 extra_info
ACTGATCGATCG...
>L2 sample_0002
ACTGATCGATCG...

Labels: L4, L2.

Requirements:

  • At least 2 distinct classes
  • At least a few sequences per class (10+ recommended for reliable training)

Marker TSV (classify and split-fastq)

Tab-separated, no header required (lines starting with # are skipped).

Columns

Column Content Required
1 Genomic position (1-based) ✅
2 REF allele ✅
3 ALT allele ✅
4+ Lineage hierarchy (one level per column) ✅ (at least one)
after first empty Annotation columns (gene, mutation) Optional

Example

#pos    ref    alt    level1    level2           gene         mutation
761155  C      T      L4        L4.9             gyrA         Ser95Thr
2155168 G      A      L2        L2.2             katG         Ser315Thr
4247431 CC     C      L1        L1.1             ethA         frameshift

Notes

  • Lineage columns are read left-to-right until the first empty cell
  • Hierarchical nesting: L4 → L4;L4.9 internally (semicolon-joined path)
  • classify works best with SNP markers (single-base REF and ALT)
  • split-fastq handles SNPs, MNVs, and small indels. However, indels in FASTQ mode are intentionally skipped because short reads produce unreliable k-mer matches across repetitive regions

Sample List TSV (classify --input-list)

sample_name    /absolute/path/to/sample.fasta    /optional/path/to/sample.gff3
  • Column 1: sample name (must be unique)
  • Column 2: path to FASTA file
  • Column 3: optional path to GFF3 annotation

Sample List TSV (split-fastq --input-list and match --input-list)

sample_name    /path/to/reads_R1.fastq.gz    /path/to/reads_R2.fastq.gz
  • Column 1: sample name (must be unique)
  • Column 2+: paths to FASTQ files (one for single-end, two for paired-end)

Reference FASTA

For classify and split-fastq: a single-record FASTA file containing the reference genome used to define marker positions.

For match: a multi-record FASTA file containing all candidate reference genomes.

GFF3 Annotation (classify)

Standard GFF3 format. Only CDS features are used. Gene names are extracted from attributes in this priority order:

  1. gene=...
  2. locus_tag=...
  3. Name=...
  4. ID=...

Coordinates are 1-based (GFF3 standard) and converted internally to 0-based for interval tree queries.