catomatic

catomatic is a Python toolkit for algorithmically constructing antimicrobial resistance (AMR) mutation catalogues directly from variant calls generated by read mapping. Rather than relying on alignment-level pattern matching or predefined resistance motifs, the tool infers resistance associations statistically from observed genotype–phenotype relationships, supporting both binary frequentist and regression-based modelling approaches.

This design is particularly well suited to Mycobacterium species, where resistance is primarily driven by chromosomal point mutations, indels, and complex multi-locus interactions, and where horizontal gene transfer is rare. By operating on mapped mutation data rather than alignment outputs, the framework enables transparent evidence tracking, flexible statistical testing, and reproducible catalogue construction tailored to the evolutionary and genomic characteristics of mycobacteria.

For aligment-relevant approaches, see AMRverse.

Introduction

This repo contains 2 approaches to build resistance catalogues:

1. Definite defectives (solo-based approach)
2. Interval regression

The first is used in https://doi.org/10.1101/2025.01.30.635633, and the second is a Python translation of the method used in https://doi.org/10.1038/s41467-023-44325-5.

---

Binary Builder

This method relies on the logic that mutations that do not cause resistance can co-occur with those that do. If a mutation in isolation (solo) does not cause resistance, then it is not contributing to the phenotype when present in a mixture either.

Mutations that occur in isolation across specified genes are traversed in sequence, and if their proportion of drug-susceptibility (vs. resistance) passes the specified statistical test, they are characterized as benign (susceptible) and removed from the dataset. This step repeats iteratively until no more solo benign mutations are found.

Remaining mutations are classified based on their resistance rates and statistical test results. Those that don’t meet thresholds are labeled as U.

The classification approach supports:

• No test: assumes homogeneous susceptibility is sufficient for S
• Binomial test: against a specified background resistance rate
• Fisher's test: using a contingency background

Optional Interventions

1. Seeding: You can pre-seed the catalogue with known neutral mutations.
2. Overrides: You can override or supplement the final catalogue with manual or rule-based entries.

Because the method uses GARC1 grammar, rules like {rpoB@*_fs:R} can be supplied post-hoc. These rules can:

• Be additive (lower priority than specific entries)
• Replace matching mutations (requires replace=True and wildcards supplied)

The catalogue can be returned as a dictionary or a Piezo-compatible pandas.DataFrame.

Contingency tables, proportions, p-values, and Wilson confidence intervals are stored in the EVIDENCE field.

Example Workflow using Fisher's:

---

Regression Builder

The Regression Builder implements a mixed-effect interval regression-based approach for catalogue construction to generate predicted mean MICs. It is suitable when the phenotypes are censored or uncesnored MICs.

If whole genome SNPs are provided, agglomerative clustering can compute random effects to control for population structure. Any given number of fixed-effects (such as lineage and lab) can also be defined by supplying additional input columns.

Similarly to the BinaryBuilder, catalogues can be exported as json objects or piezo-compatible tables.

---

Installation

Installation from source

Assuming in project directory (after git cloning)

pip install -e .

Pypy installation

pip install catomatic

Running catomatic's Binary Builder

You need two input DataFrames:

• Samples: one row per sample, with 'R' or 'S' phenotypes (UNIQUEID, PHENOTYPE)
• Mutations: one row per mutation per sample (UNIQUEID, MUTATION)

If exporting to Piezo format (--to_piezo):

• The MUTATION column must follow GARC1 grammar (gene@mutation)
• A path to a wildcards.json file (containing mutation rules) must be provided

Python/Jupyter Example

from catomatic.BinaryCatalogue import BinaryBuilder

# Build catalogue
catalogue = BinaryBuilder(samples=samples_df, mutations=mutations_df).build()

# View dictionary version
cat_dict = catalogue.return_catalogue()

# Convert to Piezo-compatible format
catalogue_df = catalogue.build_piezo(
    genbank_ref='...',
    catalogue_name='...',
    version='...',
    drug='...',
    wildcards='path/to/wildcards.json'
)

# Optionally export to CSV
catalogue.to_piezo(
    genbank_ref='...',
    catalogue_name='...',
    version='...',
    drug='...',
    wildcards='path/to/wildcards.json',
    outfile='path/to/output.csv'
)

CLI

After installation, the simplest way to run the catomatic catalogue builder is via the command line interface using the binary subcommand. You must use either --to_piezo or --to_json to specify the output format. Additional metadata is required when using --to_piezo.

Export to JSON

catomatic binary \
  --samples path/to/samples.csv \
  --mutations path/to/mutations.csv \
  --to_json \
  --outfile path/to/output/catalogue.json

Export to Piezo format

catomatic binary \
  --samples path/to/samples.csv \
  --mutations path/to/mutations.csv \
  --to_piezo \
  --outfile path/to/output/catalogue.csv \
  --genbank_ref '...' \
  --catalogue_name '...' \
  --version '...' \
  --drug '...' \
  --wildcards path/to/wildcards.json

CLI Parameters

Parameter	Type	Description
--samples	str	Path to the samples (phenotypes) file. Required.
--mutations	str	Path to the mutations file. Required.
--outfile	str	Output file path for saving the catalogue. Required with --to_json or --to_piezo.
--to_json	flag	Export the resulting catalogue in JSON format. Optional.
--to_piezo	flag	Export the resulting catalogue in Piezo-compatible CSV format. Optional.
--genbank_ref	str	GenBank reference string for Piezo export. Required with --to_piezo.
--catalogue_name	str	Name of the catalogue. Required with --to_piezo.
--version	str	Catalogue version. Required with --to_piezo.
--drug	str	Name of the drug. Required with --to_piezo.
--wildcards	str	Path to JSON file containing wildcard mutation definitions. Required with --to_piezo.
--test	str	Type of statistical test to apply. One of: None, Binomial, or Fisher. Optional.
--background	float	Background mutation rate (0–1). Required if --test Binomial is used.
--p	float	P-value threshold for statistical significance. Optional. Defaults to 0.95.
--tails	str	Tail type for statistical test. One of: one, two. Optional. Defaults to two.
--strict_unlock	flag	If set, disables classification of susceptible (S) mutations unless statistically confident.

Running catomatic's Regression Builder

You need two input DataFrames:

• Samples: one row per sample, with an MIC column (UNIQUEID, MIC)
• Mutations: one row per mutation per sample (UNIQUEID, MUTATION)

If exporting to Piezo format (--to_piezo):

• The MUTATION column must follow GARC1 grammar (gene@mutation)
• A path to a wildcards.json file (containing mutation rules) must be provided

Python/Jupyter Example

from catomatic.RegressionCatalogue import RegressionBuilder

# fit the model to generate mutation effects
model, effects = RegressionBuilder(samples=samples_df, mutations=mutations_df).predict_effects()

# classify effects and generate a catalogue (requires an ecoff)
catalogue = RegressionBuilder(samples=samples_df, mutations=mutations_df).build(ecoff=1.0)

# View dictionary version
cat_dict = catalogue.return_catalogue()

# Convert to Piezo-compatible format
catalogue_df = catalogue.build_piezo(
    genbank_ref='...',
    catalogue_name='...',
    version='...',
    drug='...',
    wildcards='path/to/wildcards.json'
)

# Optionally export to CSV
catalogue.to_piezo(
    genbank_ref='...',
    catalogue_name='...',
    version='...',
    drug='...',
    wildcards='path/to/wildcards.json',
    outfile='path/to/output.csv'
)

CLI

Similarly to BinaryBuilder, one can instantiate RegressionBuilder from the command line:

Export to JSON

catomatic regression \
  --samples path/to/samples.csv \
  --mutations path/to/mutations.csv \
  --ecoff 1.0 \
  --to_json \
  --outfile path/to/output/catalogue.json

Export to Piezo format

catomatic regression \
  --samples path/to/samples.csv \
  --mutations path/to/mutations.csv \
  --ecoff 1.0 \
  --to_piezo \
  --outfile path/to/output/catalogue.csv \
  --genbank_ref '...' \
  --catalogue_name '...' \
  --version '...' \
  --drug '...' \
  --wildcards path/to/wildcards.json

CLI Parameters

CLI Parameters (Regression Builder)

Parameter	Type	Description & default
--samples	str	Path to the samples file (CSV). Required.
--mutations	str	Path to the mutations file (CSV). Required.
--genes	str[]	List of RAV genes. Required when non-RAV genes appear in the mutations table (e.g. when clustering SNP distances). Default: [].
--dilution_factor	int	Dilution factor used in processing. Default: 2.
--censored	flag	Treat phenotype data as censored. Default: False.
--tail_dilutions	int	Tail dilutions to use for uncensored data. Default: 1.
--frs	float	Fraction Read Support threshold. Default: None.
--ecoff	float	Epidemiological cutoff value for classification. If None, it will be computed. Default: None.
--b_bounds	float,float	Bounds for beta (fixed-effect) coefficients. Two floats: (min max). Default: (None, None).
--u_bounds	float,float	Bounds for random-effect coefficients. Two floats: (min max). Default: (None, None).
--s_bounds	float,float	Bounds for sigma (residual variance). Two floats: (min max). Default: (None, None).
--p	float	Significance / confidence level. Default: 0.95.
--fixed_effects	str[]	Column names to include as fixed effects. Default: None.
--random_effects	flag	Perform SNP clustering and include cluster as a random effect. Default: False.
--cluster_distance	float	Distance threshold for SNP clustering. Default: 1.
--outfile	str	Path to save output JSON or Piezo file. Required with --to_json or --to_piezo.
--options	dict	Options passed to scipy.optimize.minimize. Default: None.
--L2_penalties	dict	Regularisation penalties for fixed and random effects. Default: None.
--to_json	flag	Export the resulting catalogue to JSON format.
--to_piezo	flag	Export the resulting catalogue to Piezo-compatible CSV format.
--genbank_ref	str	GenBank reference string for Piezo export. Required with --to_piezo.
--catalogue_name	str	Name of the catalogue. Required with --to_piezo.
--version	str	Catalogue version. Required with --to_piezo.
--drug	str	Drug associated with the mutations. Required with --to_piezo.
--wildcards	str	Path to JSON file containing wildcard mutation rules. Required with --to_piezo.
--grammar	str	Grammar used in the catalogue. Default: GARC1.
--values	str	Values used for predictions in the catalogue. Default: RUS.
--for_piezo	flag	If set, enables Piezo-specific placeholder rows. Omit if not exporting to Piezo. Default: False.

Notes

• When using post-hoc rule updates via .update(), you must provide wildcards and set replace=True if you intend to override existing entries.
• For Piezo export, placeholder entries are inserted automatically if needed to satisfy parser requirements (R, S, and U must be represented).
• The EVIDENCE column includes contingency tables, proportions, confidence intervals, and p-values, and may optionally include sample IDs if record_ids=True.
• To build a catalogue with the regression builder, as currently implemented, requires an ecoff as it will compare the predited effected against the background to supply an R/S/U label
  • To only calculate predicted effects, this can be done in Python by calling RegressionBuilder.predict_effects()

Citation

If you use catomatic in your research, please cite:

• https://doi.org/10.1099/mgen.0.001429