smftools 0.2.5__py3-none-any.whl → 0.3.0__py3-none-any.whl

smftools/__init__.py

@@ -1,20 +1,52 @@
 """smftools"""
 
+from __future__ import annotations
+
 import logging
 import warnings
+from importlib import import_module
 from importlib.metadata import version
+from typing import TYPE_CHECKING
 
-from . import cli, config, datasets, hmm
-from . import informatics as inform
-from . import machine_learning as ml
-from . import plotting as pl
-from . import preprocessing as pp
-from . import tools as tl
-from .readwrite import adata_to_df, merge_barcoded_anndatas_core, safe_read_h5ad, safe_write_h5ad
+from .readwrite import adata_to_df, safe_read_h5ad, safe_write_h5ad
 
 package_name = "smftools"
 __version__ = version(package_name)
 
+if TYPE_CHECKING:
+    from smftools import (
+        cli,
+        config,
+        datasets,
+        hmm,
+        informatics,
+        machine_learning,
+        plotting,
+        preprocessing,
+        tools,
+    )
+
+_LAZY_MODULES = {
+    "cli": "smftools.cli",
+    "config": "smftools.config",
+    "datasets": "smftools.datasets",
+    "hmm": "smftools.hmm",
+    "inform": "smftools.informatics",
+    "ml": "smftools.machine_learning",
+    "pl": "smftools.plotting",
+    "pp": "smftools.preprocessing",
+    "tl": "smftools.tools",
+}
+
+
+def __getattr__(name: str):
+    if name in _LAZY_MODULES:
+        module = import_module(_LAZY_MODULES[name])
+        globals()[name] = module
+        return module
+    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")
+
+
 __all__ = [
     "adata_to_df",
     "inform",

smftools/_settings.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from pathlib import Path
 from typing import Union
 

smftools/_version.py

@@ -1 +1,3 @@
-__version__ = "0.2.5"
+from __future__ import annotations
+
+__version__ = "0.3.0"

smftools/cli/__init__.py

@@ -0,0 +1 @@
+from __future__ import annotations

smftools/cli/archived/cli_flows.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 def flow_I(config_path):
     """
     High-level function to call for converting raw sequencing data to an adata object. 

smftools/cli/helpers.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from dataclasses import dataclass
 from pathlib import Path
 

smftools/cli/hmm_adata.py

@@ -3,18 +3,23 @@ from __future__ import annotations
 import copy
 from dataclasses import dataclass
 from pathlib import Path
-from typing import Any, List, Optional, Sequence, Tuple, Union
+from typing import TYPE_CHECKING, Any, List, Optional, Sequence, Tuple, Union
 
 import numpy as np
-import torch
 
 from smftools.logging_utils import get_logger
+from smftools.optional_imports import require
 
 # FIX: import _to_dense_np to avoid NameError
 from ..hmm.HMM import _safe_int_coords, _to_dense_np, create_hmm, normalize_hmm_feature_sets
 
 logger = get_logger(__name__)
 
+if TYPE_CHECKING:
+    import torch as torch_types
+
+torch = require("torch", extra="torch", purpose="HMM CLI")
+
 # =============================================================================
 # Helpers: extracting training arrays
 # =============================================================================

smftools/cli/load_adata.py

@@ -1,7 +1,11 @@
+from __future__ import annotations
+
 import shutil
 from pathlib import Path
 from typing import Iterable, Union
 
+import numpy as np
+
 from smftools.logging_utils import get_logger
 
 from .helpers import AdataPaths
@@ -76,6 +80,96 @@ def delete_tsvs(
                         logger.warning(f"[error] failed to remove tmp dir {td}: {e}")
 
 
+def load_adata(config_path: str):
+    """
+    CLI-facing wrapper for the load pipeline.
+
+    - Reads config CSV into ExperimentConfig
+    - Computes canonical paths for all downstream AnnData stages
+    - Registers those in the summary CSV
+    - Applies stage-skipping logic (hmm > spatial > pp_dedup > pp > raw)
+    - If needed, calls the core pipeline to actually build the raw AnnData
+
+    Returns
+    -------
+    adata : anndata.AnnData | None
+        Newly created AnnData object, or None if we skipped because a later-stage
+        AnnData already exists.
+    adata_path : pathlib.Path
+        Path to the "current" AnnData that should be used downstream.
+    cfg : ExperimentConfig
+        Config object for downstream steps.
+    """
+    from datetime import datetime
+    from importlib import resources
+
+    from ..config import ExperimentConfig, LoadExperimentConfig
+    from ..readwrite import add_or_update_column_in_csv, make_dirs
+    from .helpers import get_adata_paths
+
+    date_str = datetime.today().strftime("%y%m%d")
+
+    # -----------------------------
+    # 1) Load config into cfg
+    # -----------------------------
+    loader = LoadExperimentConfig(config_path)
+    defaults_dir = resources.files("smftools").joinpath("config")
+    cfg, report = ExperimentConfig.from_var_dict(
+        loader.var_dict, date_str=date_str, defaults_dir=defaults_dir
+    )
+
+    # Ensure base output dir
+    make_dirs([cfg.output_directory])
+
+    # -----------------------------
+    # 2) Compute and register paths
+    # -----------------------------
+    paths = get_adata_paths(cfg)
+
+    # experiment-level metadata in summary CSV
+    add_or_update_column_in_csv(cfg.summary_file, "experiment_name", cfg.experiment_name)
+    add_or_update_column_in_csv(cfg.summary_file, "config_path", config_path)
+    add_or_update_column_in_csv(cfg.summary_file, "input_data_path", cfg.input_data_path)
+    add_or_update_column_in_csv(cfg.summary_file, "input_files", [cfg.input_files])
+
+    # AnnData stage paths
+    add_or_update_column_in_csv(cfg.summary_file, "load_adata", paths.raw)
+    add_or_update_column_in_csv(cfg.summary_file, "pp_adata", paths.pp)
+    add_or_update_column_in_csv(cfg.summary_file, "pp_dedup_adata", paths.pp_dedup)
+    add_or_update_column_in_csv(cfg.summary_file, "spatial_adata", paths.spatial)
+    add_or_update_column_in_csv(cfg.summary_file, "hmm_adata", paths.hmm)
+
+    # -----------------------------
+    # 3) Stage skipping logic
+    # -----------------------------
+    if not getattr(cfg, "force_redo_load_adata", False):
+        if paths.hmm.exists():
+            logger.debug(f"HMM AnnData already exists: {paths.hmm}\nSkipping smftools load")
+            return None, paths.hmm, cfg
+        if paths.spatial.exists():
+            logger.debug(f"Spatial AnnData already exists: {paths.spatial}\nSkipping smftools load")
+            return None, paths.spatial, cfg
+        if paths.pp_dedup.exists():
+            logger.debug(
+                f"Preprocessed deduplicated AnnData already exists: {paths.pp_dedup}\n"
+                f"Skipping smftools load"
+            )
+            return None, paths.pp_dedup, cfg
+        if paths.pp.exists():
+            logger.debug(f"Preprocessed AnnData already exists: {paths.pp}\nSkipping smftools load")
+            return None, paths.pp, cfg
+        if paths.raw.exists():
+            logger.debug(
+                f"Raw AnnData from smftools load already exists: {paths.raw}\nSkipping smftools load"
+            )
+            return None, paths.raw, cfg
+
+    # If we get here, we actually want to run the full load pipeline
+    adata, adata_path, cfg = load_adata_core(cfg, paths, config_path=config_path)
+
+    return adata, adata_path, cfg
+
+
 def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
     """
     Core load pipeline.
@@ -105,9 +199,6 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
     cfg : ExperimentConfig
         (Same object, possibly with some fields updated, e.g. fasta path.)
     """
-    from pathlib import Path
-
-    import numpy as np
 
     from ..informatics.bam_functions import (
         align_and_sort_BAM,
@@ -219,6 +310,7 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
                 rg_sample_field=None,
                 progress=False,
                 auto_pair=cfg.fastq_auto_pairing,
+                samtools_backend=cfg.samtools_backend,
             )
 
             logger.info(f"Found the following barcodes in FASTQ inputs: {summary['barcodes']}")
@@ -384,7 +476,14 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
         else:
             logger.info("Making bed files from the aligned and sorted BAM file")
             aligned_BAM_to_bed(
-                aligned_sorted_output, cfg.output_directory, fasta, cfg.make_bigwigs, cfg.threads
+                aligned_sorted_output,
+                cfg.output_directory,
+                fasta,
+                cfg.make_bigwigs,
+                cfg.threads,
+                samtools_backend=cfg.samtools_backend,
+                bedtools_backend=cfg.bedtools_backend,
+                bigwig_backend=cfg.bigwig_backend,
             )
     ########################################################################################################################
 
@@ -404,7 +503,12 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
         else:
             make_dirs([cfg.split_path])
             logger.info("Demultiplexing samples into individual aligned/sorted BAM files")
-            all_bam_files = split_and_index_BAM(aligned_sorted_BAM, cfg.split_path, cfg.bam_suffix)
+            all_bam_files = split_and_index_BAM(
+                aligned_sorted_BAM,
+                cfg.split_path,
+                cfg.bam_suffix,
+                samtools_backend=cfg.samtools_backend,
+            )
 
             unclassified_bams = [p for p in all_bam_files if "unclassified" in p.name]
             bam_files = sorted(p for p in all_bam_files if "unclassified" not in p.name)
@@ -489,7 +593,16 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
         else:
             logger.info("Making BED files from BAM files for each sample")
             for bam in bam_files:
-                aligned_BAM_to_bed(bam, cfg.split_path, fasta, cfg.make_bigwigs, cfg.threads)
+                aligned_BAM_to_bed(
+                    bam,
+                    cfg.split_path,
+                    fasta,
+                    cfg.make_bigwigs,
+                    cfg.threads,
+                    samtools_backend=cfg.samtools_backend,
+                    bedtools_backend=cfg.bedtools_backend,
+                    bigwig_backend=cfg.bigwig_backend,
+                )
     ########################################################################################################################
 
     ################################### 6) SAMTools based BAM QC ######################################################################
@@ -501,7 +614,13 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
     else:
         make_dirs([bam_qc_dir])
         logger.info("Performing BAM QC")
-        bam_qc(bam_files, bam_qc_dir, cfg.threads, modality=cfg.smf_modality)
+        bam_qc(
+            bam_files,
+            bam_qc_dir,
+            cfg.threads,
+            modality=cfg.smf_modality,
+            samtools_backend=cfg.samtools_backend,
+        )
     ########################################################################################################################
 
     ################################### 7) AnnData loading ######################################################################
@@ -529,6 +648,7 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
             deaminase_footprinting,
             delete_intermediates=cfg.delete_intermediate_hdfs,
             double_barcoded_path=double_barcoded_path,
+            samtools_backend=cfg.samtools_backend,
         )
     else:
         if mod_bed_dir.is_dir():
@@ -584,6 +704,7 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
             cfg.delete_batch_hdfs,
             cfg.threads,
             double_barcoded_path,
+            cfg.samtools_backend,
         )
         if cfg.delete_intermediate_tsvs:
             delete_tsvs(mod_tsv_dir)
@@ -604,6 +725,7 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
         extract_read_features_from_bam_callable=extract_read_features_from_bam,
         bypass=cfg.bypass_add_read_length_and_mapping_qc,
         force_redo=cfg.force_redo_add_read_length_and_mapping_qc,
+        samtools_backend=cfg.samtools_backend,
     )
 
     raw_adata.obs["Raw_modification_signal"] = np.nansum(raw_adata.X, axis=1)
@@ -639,7 +761,7 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
     # multiqc ###
     mqc_dir = cfg.split_path / "multiqc"
     if mqc_dir.is_dir():
-        logger.debug(f"{mqc_dir} already exists, skipping multiqc")
+        logger.info(f"{mqc_dir} already exists, skipping multiqc")
     else:
         logger.info("Running multiqc")
         run_multiqc(cfg.split_path, mqc_dir)
@@ -665,93 +787,3 @@ def load_adata_core(cfg, paths: AdataPaths, config_path: str | None = None):
     ########################################################################################################################
 
     return raw_adata, raw_adata_path, cfg
-
-
-def load_adata(config_path: str):
-    """
-    CLI-facing wrapper for the load pipeline.
-
-    - Reads config CSV into ExperimentConfig
-    - Computes canonical paths for all downstream AnnData stages
-    - Registers those in the summary CSV
-    - Applies stage-skipping logic (hmm > spatial > pp_dedup > pp > raw)
-    - If needed, calls the core pipeline to actually build the raw AnnData
-
-    Returns
-    -------
-    adata : anndata.AnnData | None
-        Newly created AnnData object, or None if we skipped because a later-stage
-        AnnData already exists.
-    adata_path : pathlib.Path
-        Path to the "current" AnnData that should be used downstream.
-    cfg : ExperimentConfig
-        Config object for downstream steps.
-    """
-    from datetime import datetime
-    from importlib import resources
-
-    from ..config import ExperimentConfig, LoadExperimentConfig
-    from ..readwrite import add_or_update_column_in_csv, make_dirs
-    from .helpers import get_adata_paths
-
-    date_str = datetime.today().strftime("%y%m%d")
-
-    # -----------------------------
-    # 1) Load config into cfg
-    # -----------------------------
-    loader = LoadExperimentConfig(config_path)
-    defaults_dir = resources.files("smftools").joinpath("config")
-    cfg, report = ExperimentConfig.from_var_dict(
-        loader.var_dict, date_str=date_str, defaults_dir=defaults_dir
-    )
-
-    # Ensure base output dir
-    make_dirs([cfg.output_directory])
-
-    # -----------------------------
-    # 2) Compute and register paths
-    # -----------------------------
-    paths = get_adata_paths(cfg)
-
-    # experiment-level metadata in summary CSV
-    add_or_update_column_in_csv(cfg.summary_file, "experiment_name", cfg.experiment_name)
-    add_or_update_column_in_csv(cfg.summary_file, "config_path", config_path)
-    add_or_update_column_in_csv(cfg.summary_file, "input_data_path", cfg.input_data_path)
-    add_or_update_column_in_csv(cfg.summary_file, "input_files", [cfg.input_files])
-
-    # AnnData stage paths
-    add_or_update_column_in_csv(cfg.summary_file, "load_adata", paths.raw)
-    add_or_update_column_in_csv(cfg.summary_file, "pp_adata", paths.pp)
-    add_or_update_column_in_csv(cfg.summary_file, "pp_dedup_adata", paths.pp_dedup)
-    add_or_update_column_in_csv(cfg.summary_file, "spatial_adata", paths.spatial)
-    add_or_update_column_in_csv(cfg.summary_file, "hmm_adata", paths.hmm)
-
-    # -----------------------------
-    # 3) Stage skipping logic
-    # -----------------------------
-    if not getattr(cfg, "force_redo_load_adata", False):
-        if paths.hmm.exists():
-            logger.debug(f"HMM AnnData already exists: {paths.hmm}\nSkipping smftools load")
-            return None, paths.hmm, cfg
-        if paths.spatial.exists():
-            logger.debug(f"Spatial AnnData already exists: {paths.spatial}\nSkipping smftools load")
-            return None, paths.spatial, cfg
-        if paths.pp_dedup.exists():
-            logger.debug(
-                f"Preprocessed deduplicated AnnData already exists: {paths.pp_dedup}\n"
-                f"Skipping smftools load"
-            )
-            return None, paths.pp_dedup, cfg
-        if paths.pp.exists():
-            logger.debug(f"Preprocessed AnnData already exists: {paths.pp}\nSkipping smftools load")
-            return None, paths.pp, cfg
-        if paths.raw.exists():
-            logger.debug(
-                f"Raw AnnData from smftools load already exists: {paths.raw}\nSkipping smftools load"
-            )
-            return None, paths.raw, cfg
-
-    # If we get here, we actually want to run the full load pipeline
-    adata, adata_path, cfg = load_adata_core(cfg, paths, config_path=config_path)
-
-    return adata, adata_path, cfg

smftools/cli/preprocess_adata.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from pathlib import Path
 from typing import Optional, Tuple
 

smftools/cli/spatial_adata.py

@@ -1,9 +1,12 @@
+from __future__ import annotations
+
 from pathlib import Path
 from typing import Optional, Tuple
 
 import anndata as ad
 
 from smftools.logging_utils import get_logger
+from smftools.optional_imports import require
 
 logger = get_logger(__name__)
 
@@ -153,7 +156,8 @@ def spatial_adata_core(
 
     import numpy as np
     import pandas as pd
-    import scanpy as sc
+
+    sc = require("scanpy", extra="scanpy", purpose="spatial analyses")
 
     from ..metadata import record_smftools_metadata
     from ..plotting import (

smftools/cli_entry.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import logging
 from pathlib import Path
 from typing import Sequence
@@ -10,10 +12,32 @@ from .cli.load_adata import load_adata
 from .cli.preprocess_adata import preprocess_adata
 from .cli.spatial_adata import spatial_adata
 from .informatics.pod5_functions import subsample_pod5
-from .logging_utils import setup_logging
+from .logging_utils import get_logger, setup_logging
 from .readwrite import concatenate_h5ads
 
 
+def _configure_multiprocessing() -> None:
+    import multiprocessing as mp
+    import sys
+
+    logger = get_logger(__name__)
+
+    try:
+        if sys.platform == "win32":
+            mp.set_start_method("spawn")
+            logger.debug("Setting multiprocessing start method to spawn")
+        else:
+            # try forkserver first, fallback to spawn
+            try:
+                mp.set_start_method("forkserver")
+                logger.debug("Setting multiprocessing start method to forkserver")
+            except ValueError:
+                mp.set_start_method("spawn")
+                logger.debug("Setting multiprocessing start method to spawn")
+    except RuntimeError:
+        logger.warning("Could not set multiprocessing start method")
+
+
 @click.group()
 @click.option(
     "--log-file",
@@ -32,6 +56,7 @@ def cli(log_file: Path | None, log_level: str):
     """Command-line interface for smftools."""
     level = getattr(logging, log_level.upper(), logging.INFO)
     setup_logging(level=level, log_file=log_file)
+    _configure_multiprocessing()
 
 
 ####### Load anndata from raw data ###########

smftools/config/__init__.py

@@ -1 +1,3 @@
+from __future__ import annotations
+
 from .experiment_config import ExperimentConfig, LoadExperimentConfig

smftools/config/default.yaml

@@ -77,6 +77,9 @@ aligner_args:
 # Sorted BAM and BED specific handling
 make_bigwigs: False # Whether to make coverage bigwigs
 make_beds: False # Whether to make beds from the aligned bams
+samtools_backend: auto # auto|python|cli for samtools-compatible operations
+bedtools_backend: auto # auto|python|cli for bedtools-compatible operations
+bigwig_backend: auto # auto|python|cli for bedGraphToBigWig conversion
 
 # Nanopore specific demultiplexing
 barcode_both_ends: False # dorado demultiplexing
@@ -370,4 +373,4 @@ force_redo_matrix_corr_plotting: False # Whether to force redo basic correlation
 bypass_hmm_fit: False # Whether to skip HMM fitting for each sample/reference
 force_redo_hmm_fit: False # Whether to redo HMM fitting for each sample/reference
 bypass_hmm_apply: False # Whether to skip HMM application for each sample/reference
-force_redo_hmm_apply: False # Whether to redo HMM application for each sample/reference
+force_redo_hmm_apply: False # Whether to redo HMM application for each sample/reference

smftools/config/experiment_config.py

@@ -736,6 +736,9 @@ class ExperimentConfig:
     aligner_args: Optional[List[str]] = None
     make_bigwigs: bool = False
     make_beds: bool = False
+    samtools_backend: str = "auto"
+    bedtools_backend: str = "auto"
+    bigwig_backend: str = "auto"
 
     # Anndata structure
     reference_column: Optional[str] = REF_COL
@@ -1264,6 +1267,9 @@ class ExperimentConfig:
             device=merged.get("device", "auto"),
             make_bigwigs=merged.get("make_bigwigs", False),
             make_beds=merged.get("make_beds", False),
+            samtools_backend=merged.get("samtools_backend", "auto"),
+            bedtools_backend=merged.get("bedtools_backend", "auto"),
+            bigwig_backend=merged.get("bigwig_backend", "auto"),
             delete_intermediate_hdfs=merged.get("delete_intermediate_hdfs", True),
             mod_target_bases=merged.get("mod_target_bases", ["GpC", "CpG"]),
             enzyme_target_bases=merged.get("enzyme_target_bases", ["GpC"]),

smftools/datasets/__init__.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from .datasets import Kissiov_and_McKenna_2025, dCas9_kinetics
 
 __all__ = ["dCas9_kinetics", "Kissiov_and_McKenna_2025"]

smftools/hmm/HMM.py

@@ -3,14 +3,20 @@ from __future__ import annotations
 import ast
 import json
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Sequence, Tuple, Union
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, Sequence, Tuple, Union
 
 import numpy as np
-import torch
-import torch.nn as nn
 from scipy.sparse import issparse
 
 from smftools.logging_utils import get_logger
+from smftools.optional_imports import require
+
+if TYPE_CHECKING:
+    import torch as torch_types
+    import torch.nn as nn_types
+
+torch = require("torch", extra="torch", purpose="HMM modeling")
+nn = torch.nn
 
 logger = get_logger(__name__)
 # =============================================================================

smftools/hmm/__init__.py

@@ -1,13 +1,24 @@
-from .call_hmm_peaks import call_hmm_peaks
-from .display_hmm import display_hmm
-from .hmm_readwrite import load_hmm, save_hmm
-from .nucleosome_hmm_refinement import infer_nucleosomes_in_large_bound, refine_nucleosome_calls
-
-__all__ = [
-    "call_hmm_peaks",
-    "display_hmm",
-    "load_hmm",
-    "refine_nucleosome_calls",
-    "infer_nucleosomes_in_large_bound",
-    "save_hmm",
-]
+from __future__ import annotations
+
+from importlib import import_module
+
+_LAZY_ATTRS = {
+    "call_hmm_peaks": "smftools.hmm.call_hmm_peaks",
+    "display_hmm": "smftools.hmm.display_hmm",
+    "load_hmm": "smftools.hmm.hmm_readwrite",
+    "save_hmm": "smftools.hmm.hmm_readwrite",
+    "infer_nucleosomes_in_large_bound": "smftools.hmm.nucleosome_hmm_refinement",
+    "refine_nucleosome_calls": "smftools.hmm.nucleosome_hmm_refinement",
+}
+
+
+def __getattr__(name: str):
+    if name in _LAZY_ATTRS:
+        module = import_module(_LAZY_ATTRS[name])
+        attr = getattr(module, name)
+        globals()[name] = attr
+        return attr
+    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")
+
+
+__all__ = list(_LAZY_ATTRS.keys())

smftools/hmm/archived/apply_hmm_batched.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import numpy as np
 import pandas as pd
 import torch

smftools/hmm/archived/calculate_distances.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 # calculate_distances
 
 def calculate_distances(intervals, threshold=0.9):

smftools/hmm/archived/call_hmm_peaks.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 def call_hmm_peaks(
     adata,
     feature_configs,

smftools/hmm/archived/train_hmm.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 def train_hmm(
     data,
     emission_probs=[[0.8, 0.2], [0.2, 0.8]],

smftools/hmm/call_hmm_peaks.py

@@ -1,9 +1,11 @@
-# FILE: smftools/hmm/call_hmm_peaks.py
+from __future__ import annotations
 
+# FILE: smftools/hmm/call_hmm_peaks.py
 from pathlib import Path
 from typing import Any, Dict, Optional, Sequence, Union
 
 from smftools.logging_utils import get_logger
+from smftools.optional_imports import require
 
 logger = get_logger(__name__)
 
@@ -35,12 +37,13 @@ def call_hmm_peaks(
       - adata.var["is_in_any_{layer}_peak_{ref}"]
       - adata.var["is_in_any_peak"] (global)
     """
-    import matplotlib.pyplot as plt
     import numpy as np
     import pandas as pd
     from scipy.signal import find_peaks
     from scipy.sparse import issparse
 
+    plt = require("matplotlib.pyplot", extra="plotting", purpose="HMM peak plots")
+
     if not inplace:
         adata = adata.copy()