dmsroute 0.1.2__py3-none-any.whl

dmsroute/__init__.py

@@ -0,0 +1,249 @@
+"""Public API for dmsroute."""
+
+from __future__ import annotations
+
+import logging
+
+from dmsroute._version import __version__
+from dmsroute.builders import (
+    build_mavedb_dataset,
+    build_proteingym_dataset,
+    build_proteingym_indel_dataset,
+)
+from dmsroute.acquisition.cache import CacheManifest, FilesystemCache
+from dmsroute.acquisition.cache_inventory import (
+    CacheInventoryArtifact,
+    CacheInventoryEntry,
+    CacheInventoryIssue,
+    CacheInventoryResult,
+    inventory_cache,
+)
+from dmsroute.catalog import DatasetRecord, get_dataset_metadata, list_datasets
+from dmsroute.config import load_pipeline_config, validate_pipeline_config
+from dmsroute.downloads import (
+    DatasetBatchDownloadEntry,
+    DatasetBatchDownloadResult,
+    DatasetDownloadResult,
+    download_and_standardize_dataset,
+    download_and_standardize_datasets,
+)
+from dmsroute.core.exceptions import (
+    CacheError,
+    CacheInventoryError,
+    CatalogError,
+    CorruptCacheManifestError,
+    DMSParserError,
+    DatasetNotFoundError,
+    DownloadError,
+    FileFormatError,
+    InvalidCacheEntryError,
+    InvalidCatalogQueryError,
+    InvalidDatasetError,
+    InvalidHGVSVariantError,
+    InvalidPipelineOptionError,
+    InvalidSnapshotSelectorError,
+    InvalidVariantError,
+    MissingWildTypeError,
+    MutationApplicationError,
+    MaveDBSnapshotError,
+    MaveDBSnapshotTableError,
+    PipelineError,
+    SequenceValidationError,
+    SourceConfigurationError,
+    SnapshotDatasetNotFoundError,
+    SupersededSnapshotDatasetError,
+    UnknownSourceResourceError,
+    UnsupportedSourceResourceError,
+    UnsupportedVariantError,
+    WildTypeConflictError,
+)
+from dmsroute.acquisition.fetch import fetch_to_cache
+from dmsroute.acquisition.io import (
+    download_file,
+    ensure_local_copy,
+    infer_filename_from_url,
+    read_table,
+    write_table,
+)
+from dmsroute.core.parsing import (
+    apply_mutations,
+    count_mutations,
+    explode_mutations,
+    extract_mutation_tokens,
+    extract_positions,
+    hgvs_pro_is_indel,
+    hgvs_to_sequence,
+    is_wildtype_variant,
+    parse_hgvs_pro,
+    parse_mavedb_hgvs_series,
+    parse_variant,
+    parse_variant_series,
+    parse_variant_token,
+    read_fasta_one,
+    translate_dna,
+    variant_to_sequence,
+)
+from dmsroute.core.transforms import (
+    add_minmax,
+    add_pseudo_binary_label,
+    add_wt_relative_score,
+    add_zscore,
+    compute_wt_score,
+)
+from dmsroute.pipeline import PipelineResult, run_pipeline
+from dmsroute.sources.mavedb_catalog import MaveDBCatalog
+from dmsroute.sources.mavedb_bulk_catalog import (
+    MaveDBBulkCatalog,
+    MaveDBDiscoveredExperiment,
+    MaveDBDiscoveredScoreSet,
+    MaveDBDiscoveryResult,
+)
+from dmsroute.sources.mavedb_snapshots import (
+    MaveDBSnapshot,
+    MaveDBSnapshotRecord,
+    fetch_mavedb_snapshot,
+    load_cached_mavedb_snapshot,
+    resolve_mavedb_snapshot,
+)
+from dmsroute.sources.mavedb_snapshot_tables import (
+    MaveDBSnapshotTable,
+    MaveDBSnapshotTableExtractionResult,
+    extract_mavedb_snapshot_tables,
+)
+from dmsroute.sources.proteingym_catalog import ProteinGymCatalog
+from dmsroute.sources.proteingym_resources import (
+    PROTEINGYM_RESOURCES,
+    ProteinGymResource,
+    get_proteingym_resource,
+    list_proteingym_resources,
+)
+from dmsroute.core.validation import (
+    count_invalid_sequences,
+    has_wildtype_row,
+    validate_column_non_empty,
+    validate_consistent_sequence_lengths,
+    validate_n_mutations_column,
+    validate_required_columns,
+    validate_score_column,
+    validate_sequence_column,
+    validate_standard_dataset,
+    validate_status_column,
+    validate_variant_column,
+    validate_wt_presence,
+    validate_wt_sequence,
+)
+
+logger = logging.getLogger(__name__)
+logger.addHandler(logging.NullHandler())
+
+__all__ = [
+    "__version__",
+    "build_mavedb_dataset",
+    "build_proteingym_dataset",
+    "build_proteingym_indel_dataset",
+    "DatasetRecord",
+    "MaveDBCatalog",
+    "MaveDBSnapshot",
+    "MaveDBSnapshotRecord",
+    "fetch_mavedb_snapshot",
+    "load_cached_mavedb_snapshot",
+    "resolve_mavedb_snapshot",
+    "MaveDBBulkCatalog",
+    "MaveDBDiscoveredExperiment",
+    "MaveDBDiscoveredScoreSet",
+    "MaveDBDiscoveryResult",
+    "MaveDBSnapshotTable",
+    "MaveDBSnapshotTableExtractionResult",
+    "extract_mavedb_snapshot_tables",
+    "ProteinGymCatalog",
+    "ProteinGymResource",
+    "PROTEINGYM_RESOURCES",
+    "get_proteingym_resource",
+    "list_proteingym_resources",
+    "list_datasets",
+    "get_dataset_metadata",
+    "load_pipeline_config",
+    "validate_pipeline_config",
+    "run_pipeline",
+    "PipelineResult",
+    "download_and_standardize_dataset",
+    "DatasetDownloadResult",
+    "download_and_standardize_datasets",
+    "DatasetBatchDownloadEntry",
+    "DatasetBatchDownloadResult",
+    "FilesystemCache",
+    "CacheManifest",
+    "CacheInventoryArtifact",
+    "CacheInventoryEntry",
+    "CacheInventoryIssue",
+    "CacheInventoryResult",
+    "CacheInventoryError",
+    "inventory_cache",
+    "fetch_to_cache",
+    "parse_variant",
+    "parse_variant_token",
+    "parse_variant_series",
+    "parse_hgvs_pro",
+    "parse_mavedb_hgvs_series",
+    "count_mutations",
+    "is_wildtype_variant",
+    "explode_mutations",
+    "extract_positions",
+    "extract_mutation_tokens",
+    "hgvs_pro_is_indel",
+    "apply_mutations",
+    "variant_to_sequence",
+    "hgvs_to_sequence",
+    "read_fasta_one",
+    "translate_dna",
+    "compute_wt_score",
+    "add_wt_relative_score",
+    "add_pseudo_binary_label",
+    "add_zscore",
+    "add_minmax",
+    "download_file",
+    "infer_filename_from_url",
+    "ensure_local_copy",
+    "read_table",
+    "write_table",
+    "validate_required_columns",
+    "validate_column_non_empty",
+    "validate_score_column",
+    "validate_variant_column",
+    "validate_status_column",
+    "validate_wt_sequence",
+    "validate_wt_presence",
+    "validate_n_mutations_column",
+    "validate_sequence_column",
+    "validate_consistent_sequence_lengths",
+    "validate_standard_dataset",
+    "has_wildtype_row",
+    "count_invalid_sequences",
+    "DMSParserError",
+    "CacheError",
+    "CatalogError",
+    "InvalidCacheEntryError",
+    "CorruptCacheManifestError",
+    "InvalidCatalogQueryError",
+    "DatasetNotFoundError",
+    "SourceConfigurationError",
+    "UnknownSourceResourceError",
+    "UnsupportedSourceResourceError",
+    "PipelineError",
+    "InvalidPipelineOptionError",
+    "MaveDBSnapshotError",
+    "MaveDBSnapshotTableError",
+    "SnapshotDatasetNotFoundError",
+    "SupersededSnapshotDatasetError",
+    "InvalidSnapshotSelectorError",
+    "InvalidDatasetError",
+    "MissingWildTypeError",
+    "WildTypeConflictError",
+    "InvalidVariantError",
+    "InvalidHGVSVariantError",
+    "UnsupportedVariantError",
+    "SequenceValidationError",
+    "MutationApplicationError",
+    "DownloadError",
+    "FileFormatError",
+]

dmsroute/_version.py

@@ -0,0 +1,3 @@
+"""Package version."""
+
+__version__ = "0.1.2"