Rule-Based Model Example

Inspect biological labels derived from learned rule cutoffs before HMM segmentation.

This example loads a packaged saved model so that the focus stays on the demonstrated operation.

Parameters

Change these explicit values for your own data or output location.

from pathlib import Path

from methylseg import MethylSegPathway, MethylStateAssignmentMethod, MethylationStates
from methylseg.helper_classes import DATA_DIR

REFERENCE_DIR = DATA_DIR / "reference_files"
OUTPUT_ROOT = Path("out")
CHROM = "chr1"

RESOLUTION = "450k"
SAMPLE_NAME = "TCGA-BD-A3EP-01A"
SAMPLE_FILE = REFERENCE_DIR / "TCGA-BD-A3EP-01A_450k.tsv.gz"
OUT_DIR = OUTPUT_ROOT / "rule_based_model_example"

Load a saved model and prepare the sample

model = MethylSegPathway.get_pretrained_model(
    out_dir=OUT_DIR,
    resolution=RESOLUTION,
)
sample_info, removed_df = MethylSegPathway.prepare_sample_info(
    sample_name=SAMPLE_NAME,
    sample_file=SAMPLE_FILE,
    resolution=RESOLUTION,
    min_coverage=10,
)
sample_info.sample_id

'TCGA-BD-A3EP-01A'

Rule-based labels and HMM regions

model.state_assignment_method = MethylStateAssignmentMethod.DEFINITION
model.segmentor.state_assignment_method = MethylStateAssignmentMethod.DEFINITION
model.analyzer.pretty_print_rules()
regions_chr1 = model.generate_regions(sample_info=sample_info, chrom=CHROM)
model.plot_labels(label_source="rule_based", sample_info=sample_info, sample_info_removed=removed_df, chrom=CHROM)
model.plot_labels(label_source="hmm", sample_info=sample_info, sample_info_removed=removed_df, chrom=CHROM)
regions_chr1.head()

PMD:
0.158 <= beta <= 0.743
AND (40kb_int_pct >= 0.616 AND 40kb_std <= 0.290 AND 40kb_high_pct <= 0.374 AND 40kb_low_pct <= 0.376) OR (450kb_int_pct >= 0.596 AND 450kb_std <= 0.199 AND 450kb_high_pct <= 0.208 AND 450kb_low_pct <= 0.249)

Low methylation:
beta <= 0.158 AND NOT PMD

Intermediate methylation:
0.158 < beta < 0.743 AND NOT PMD

High methylation:
beta >= 0.743 AND NOT PMD

	CpG_chrm	start	end	avg_beta	probe_count	state	length
0	chr1	69590	855650	0.469783	46	PMD	786060
2	chr1	860612	872386	0.518261	23	INTERMEDIATE	11774
4	chr1	877488	889216	0.859375	16	HIGH	11728
7	chr1	900423	916059	0.382414	29	INTERMEDIATE	15636
8	chr1	916497	922011	0.459375	16	PMD	5514

Outputs

Raw region BED files are written under OUT_DIR. Cleaning writes chromosome-local metadata under OUT_DIR / "clean_regions", which is then used by cleaned overlays.