pub version 0.1

kinactive/__about__.py

@@ -1,3 +1,3 @@
 __author__ = "Ivan Reveguk"
 __author_email__ = "<ivan.reveguk@gmail.com>"
-__version__ = "0.1.dev1"
+__version__ = "0.1"

kinactive/db.py

@@ -20,12 +20,10 @@
     ChainList,
     ChainSequence,
     ChainStructure,
-    read_chains,
-    ChainIOConfig,
 )
 from lXtractor.ext import PDB, PyHMMer, SIFTS, fetch_uniprot, filter_by_method
 from lXtractor.util import get_files, read_fasta, write_fasta
-from more_itertools import ilen, take, consume, unzip
+from more_itertools import ilen, consume, unzip
 from toolz import curry, groupby, itemmap, keyfilter, keymap
 from tqdm.auto import tqdm
 
@@ -53,13 +51,13 @@ def _get_remaining(names: abc.Iterable[str], dir_: Path) -> set[str]:
 
 
 def _is_sequence_of_chain_seqs(
-    s: abc.Sequence[t.Any],
+        s: abc.Sequence[t.Any],
 ) -> t.TypeGuard[abc.Sequence[ChainSequence]]:
     return all(isinstance(x, ChainSequence) for x in s)
 
 
 def _stage_chain_init(
-    seq: T, pdb_chains: abc.Iterable[str], pdb_dir: Path, fmt: str
+        seq: T, pdb_chains: abc.Iterable[str], pdb_dir: Path, fmt: str
 ) -> tuple[T, list[tuple[Path, list[str]]]]:
     id2chains = groupby(op.itemgetter(0), map(lambda x: x.split(":"), pdb_chains))
     path2chains = itemmap(
@@ -71,7 +69,7 @@ def _stage_chain_init(
 
 
 def _filter_by_size(
-    structures: list[ChainStructure], cfg: DBConfig
+        structures: list[ChainStructure], cfg: DBConfig
 ) -> list[ChainStructure]:
     return [s for s in structures if len(s.seq) >= cfg.pdb_str_min_size]
 
@@ -173,10 +171,10 @@ def _get_sifts_xray(self) -> list[str]:
         return filter_by_method(sifts.pdb_ids, pdb=pdb, method="X-ray")
 
     def build(
-        self,
-        uniprot_ids: abc.Collection[str] | None = None,
-        pdb_chain_ids: abc.Collection[str] | None = None,
-        n_domains: int = 0,
+            self,
+            uniprot_ids: abc.Collection[str] | None = None,
+            pdb_chain_ids: abc.Collection[str] | None = None,
+            n_domains: int = 0,
     ) -> ChainList[Chain]:
         """
         Build a new lXt-PK data collection.
@@ -204,8 +202,8 @@ def filter_domain_str_by_canon_seq_match(c: Chain) -> Chain:
             match_name = "Match_seq1_seq1_canonical"
             c = c.apply_structures(match_seq).filter_structures(
                 lambda s: (
-                    len(s.seq) >= self.cfg.pk_min_str_domain_size
-                    and s.seq.meta[match_name] >= self.cfg.pk_min_str_seq_match
+                        len(s.seq) >= self.cfg.pk_min_str_domain_size
+                        and s.seq.meta[match_name] >= self.cfg.pk_min_str_seq_match
                 )
             )
             return c
@@ -370,12 +368,12 @@ def filter_domain_str_by_canon_seq_match(c: Chain) -> Chain:
         return chains
 
     def save(
-        self,
-        dest: Path | None = None,
-        chains: abc.Iterable[Chain] | None = None,
-        *,
-        overwrite: bool = False,
-        summary: bool = True,
+            self,
+            dest: Path | None = None,
+            chains: abc.Iterable[Chain] | None = None,
+            *,
+            overwrite: bool = False,
+            summary: bool = True,
     ) -> None:
         """
         Save DB sequence to file system.
@@ -407,7 +405,7 @@ def save(
                 LOGGER.info(f"Saved summary file {name} to {dest}")
 
     def load(
-        self, dump: Path | abc.Iterable[Path]
+            self, dump: Path | abc.Iterable[Path]
     ) -> ChainList[Chain]:
         """
         Load prepared db.
@@ -422,12 +420,17 @@ def load(
         dump = list(dump)
         LOGGER.info(f"Got {len(dump)} initial paths to read")
 
-        chains = read_chains(
-            dump,
-            children=True,
-            seq_cfg=ChainIOConfig(verbose=self.cfg.verbose),
-            str_cfg=ChainIOConfig(verbose=self.cfg.verbose, num_proc=self.cfg.io_cpus),
+        io = ChainIO(
+            num_proc=self.cfg.io_cpus,
+            verbose=self.cfg.verbose,
+            tolerate_failures=True,
+        )
+        chain_read_it = io.read_chain(
+            dump, callbacks=[chain_tree.recover], search_children=True
         )
+
+        chains = ChainList(chain_read_it)
+
         chains = chains.apply(
             chain_tree.recover,
             verbose=self.cfg.verbose,

kinactive/distances.py

@@ -172,9 +172,9 @@ def superpose(
             to this output (inplace).
         """
         try:
-            _key = {'min': min, 'max': max}[key]
+            _key = {"min": min, "max": max}[key]
         except KeyError as e:
-            raise ValueError('Invalid key') from e
+            raise ValueError("Invalid key") from e
 
         structures = ChainList(structures)
         if len(structures) == 1: