support residue variants accepted by TALOS: protonated HIS and oxidis…

…ed CYS
varioustoxins · Nov 5, 2023 · 7b24ebb · 7b24ebb
1 parent c6338a5
commit 7b24ebb
Show file tree

Hide file tree

Showing 7 changed files with 427 additions and 8 deletions.
diff --git a/src/nef_pipelines/lib/sequence_lib.py b/src/nef_pipelines/lib/sequence_lib.py
@@ -498,6 +498,7 @@ def sequence_from_frame(
     sequence_code_index = loop.tag_index("sequence_code")
     residue_name_index = loop.tag_index("residue_name")
     linking_index = loop.tag_index("linking")
+    residue_variant_index = loop.tag_index("residue_variant")
 
     for line in loop:
         chain_code = line[chain_code_index]
@@ -512,11 +513,22 @@ def sequence_from_frame(
                 if line[linking_index] != NEF_UNKNOWN
                 else None
             )
+            residue_variants = line[residue_variant_index].split(",")
+            residue_variants = (
+                ()
+                if residue_variants
+                == [
+                    UNUSED,
+                ]
+                else tuple(residue_variants)
+            )
+
             residue = SequenceResidue(
                 chain_code=chain_code,
                 sequence_code=sequence_code,
                 residue_name=residue_name,
                 linking=linking,
+                variants=residue_variants,
             )
             residues.append(residue)
 

diff --git a/src/nef_pipelines/lib/structures.py b/src/nef_pipelines/lib/structures.py
@@ -39,7 +39,7 @@ class SequenceResidue(Residue):
 
     is_cis: bool = False
     linking: Optional[Linking] = None
-    variant: Optional[str] = None
+    variants: List[str] = ()
 
 
 # should contain a residue and have constructors?

diff --git a/src/nef_pipelines/tests/talos/test_data/oxidised_cys.nef b/src/nef_pipelines/tests/talos/test_data/oxidised_cys.nef
@@ -0,0 +1,88 @@
+data_nmrpipe
+
+save_nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_category      nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_framecode     nef_nmr_meta_data
+   _nef_nmr_meta_data.format_name      nmr_exchange_format
+   _nef_nmr_meta_data.format_version   1.1
+   _nef_nmr_meta_data.program_name     NEFPipelines
+   _nef_nmr_meta_data.script_name      nmrpipe/import/shifts.py
+   _nef_nmr_meta_data.program_version  0.1.45
+   _nef_nmr_meta_data.creation_date    2023-08-27T12:11:33.808424
+   _nef_nmr_meta_data.uuid             NEFPipelines-2023-08-27T12:11:33.930553-9145925061
+   _nef_nmr_meta_data.creation_time    2023-08-27T12:11:33.930553
+
+   loop_
+      _nef_run_history.run_number
+      _nef_run_history.program_name
+      _nef_run_history.program_version
+      _nef_run_history.script_name
+
+     1   NEFPipelines   0.1.45   nmrpipe/import/sequence.py
+
+   stop_
+
+save_
+
+save_nef_molecular_system
+   _nef_molecular_system.sf_category   nef_molecular_system
+   _nef_molecular_system.sf_framecode  nef_molecular_system
+
+   loop_
+      _nef_sequence.index
+      _nef_sequence.chain_code
+      _nef_sequence.sequence_code
+      _nef_sequence.residue_name
+      _nef_sequence.linking
+      _nef_sequence.residue_variant
+      _nef_sequence.cis_peptide
+
+     1    A   1    MET   start    .     .
+     2    A   2    CYS   middle   -HG   .
+     3    A   3    ILE   middle   .     .
+     4    A   4    PHE   end      .     .
+
+   stop_
+
+save_
+
+save_nef_chemical_shift_list_nmrpipe
+   _nef_chemical_shift_list.sf_category   nef_chemical_shift_list
+   _nef_chemical_shift_list.sf_framecode  nef_chemical_shift_list_nmrpipe
+
+   loop_
+      _nef_chemical_shift.chain_code
+      _nef_chemical_shift.sequence_code
+      _nef_chemical_shift.residue_name
+      _nef_chemical_shift.atom_name
+      _nef_chemical_shift.value
+      _nef_chemical_shift.value_uncertainty
+      _nef_chemical_shift.element
+      _nef_chemical_shift.isotope_number
+
+     A   1    MET   CB    33.27    .   .   .
+     A   1    MET   CA    54.45    .   .   .
+     A   1    MET   C     170.54   .   .   .
+     A   1    MET   HA    4.23     .   .   .
+     A   2    CYS   HN    8.9      .   .   .
+     A   2    CYS   CA    55.08    .   .   .
+     A   2    CYS   CB    30.76    .   .   .
+     A   2    CYS   C     175.92   .   .   .
+     A   2    CYS   HA    5.249    .   .   .
+     A   2    CYS   N     123.22   .   .   .
+     A   3    ILE   HA    4.213    .   .   .
+     A   3    ILE   C     172.45   .   .   .
+     A   3    ILE   CA    59.57    .   .   .
+     A   3    ILE   N     115.34   .   .   .
+     A   3    ILE   HN    8.32     .   .   .
+     A   3    ILE   CB    42.21    .   .   .
+     A   4    PHE   HA    5.63     .   .   .
+     A   4    PHE   C     175.32   .   .   .
+     A   4    PHE   CB    41.48    .   .   .
+     A   4    PHE   CA    55.21    .   .   .
+     A   4    PHE   N     118.11   .   .   .
+     A   4    PHE   HN    8.61     .   .   .
+
+   stop_
+
+save_
diff --git a/src/nef_pipelines/tests/talos/test_data/protonated_his.nef b/src/nef_pipelines/tests/talos/test_data/protonated_his.nef
@@ -0,0 +1,89 @@
+data_nmrpipe
+
+save_nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_category      nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_framecode     nef_nmr_meta_data
+   _nef_nmr_meta_data.format_name      nmr_exchange_format
+   _nef_nmr_meta_data.format_version   1.1
+   _nef_nmr_meta_data.program_name     NEFPipelines
+   _nef_nmr_meta_data.script_name      nmrpipe/import/shifts.py
+   _nef_nmr_meta_data.program_version  0.1.45
+   _nef_nmr_meta_data.creation_date    2023-08-27T12:11:33.808424
+   _nef_nmr_meta_data.uuid             NEFPipelines-2023-08-27T12:11:33.930553-9145925061
+   _nef_nmr_meta_data.creation_time    2023-08-27T12:11:33.930553
+
+   loop_
+      _nef_run_history.run_number
+      _nef_run_history.program_name
+      _nef_run_history.program_version
+      _nef_run_history.script_name
+
+     1   NEFPipelines   0.1.45   nmrpipe/import/sequence.py
+
+   stop_
+
+save_
+
+save_nef_molecular_system
+   _nef_molecular_system.sf_category   nef_molecular_system
+   _nef_molecular_system.sf_framecode  nef_molecular_system
+
+   loop_
+      _nef_sequence.index
+      _nef_sequence.chain_code
+      _nef_sequence.sequence_code
+      _nef_sequence.residue_name
+      _nef_sequence.linking
+      _nef_sequence.residue_variant
+      _nef_sequence.cis_peptide
+
+     1    A   1    MET   start    .      .
+     2    A   2    HIS   middle   +HE2   .
+     3    A   3    ILE   middle   .      .
+     4    A   4    PHE   end      .      .
+
+
+   stop_
+
+save_
+
+save_nef_chemical_shift_list_nmrpipe
+   _nef_chemical_shift_list.sf_category   nef_chemical_shift_list
+   _nef_chemical_shift_list.sf_framecode  nef_chemical_shift_list_nmrpipe
+
+   loop_
+      _nef_chemical_shift.chain_code
+      _nef_chemical_shift.sequence_code
+      _nef_chemical_shift.residue_name
+      _nef_chemical_shift.atom_name
+      _nef_chemical_shift.value
+      _nef_chemical_shift.value_uncertainty
+      _nef_chemical_shift.element
+      _nef_chemical_shift.isotope_number
+
+     A   1    MET   CB    33.27    .   .   .
+     A   1    MET   CA    54.45    .   .   .
+     A   1    MET   C     170.54   .   .   .
+     A   1    MET   HA    4.23     .   .   .
+     A   2    HIS   HN    8.9      .   .   .
+     A   2    HIS   CA    55.08    .   .   .
+     A   2    HIS   CB    30.76    .   .   .
+     A   2    HIS   C     175.92   .   .   .
+     A   2    HIS   HA    5.249    .   .   .
+     A   2    HIS   N     123.22   .   .   .
+     A   3    ILE   HA    4.213    .   .   .
+     A   3    ILE   C     172.45   .   .   .
+     A   3    ILE   CA    59.57    .   .   .
+     A   3    ILE   N     115.34   .   .   .
+     A   3    ILE   HN    8.32     .   .   .
+     A   3    ILE   CB    42.21    .   .   .
+     A   4    PHE   HA    5.63     .   .   .
+     A   4    PHE   C     175.32   .   .   .
+     A   4    PHE   CB    41.48    .   .   .
+     A   4    PHE   CA    55.21    .   .   .
+     A   4    PHE   N     118.11   .   .   .
+     A   4    PHE   HN    8.61     .   .   .
+
+   stop_
+
+save_
diff --git a/src/nef_pipelines/tests/talos/test_data/ubi_4_offset_1.nef b/src/nef_pipelines/tests/talos/test_data/ubi_4_offset_1.nef
@@ -0,0 +1,89 @@
+data_nmrpipe
+
+save_nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_category      nef_nmr_meta_data
+   _nef_nmr_meta_data.sf_framecode     nef_nmr_meta_data
+   _nef_nmr_meta_data.format_name      nmr_exchange_format
+   _nef_nmr_meta_data.format_version   1.1
+   _nef_nmr_meta_data.program_name     NEFPipelines
+   _nef_nmr_meta_data.script_name      nmrpipe/import/shifts.py
+   _nef_nmr_meta_data.program_version  0.1.45
+   _nef_nmr_meta_data.creation_date    2023-08-27T12:11:33.808424
+   _nef_nmr_meta_data.uuid             NEFPipelines-2023-08-27T12:11:33.930553-9145925061
+   _nef_nmr_meta_data.creation_time    2023-08-27T12:11:33.930553
+
+   loop_
+      _nef_run_history.run_number
+      _nef_run_history.program_name
+      _nef_run_history.program_version
+      _nef_run_history.script_name
+
+     1   NEFPipelines   0.1.45   nmrpipe/import/sequence.py
+
+   stop_
+
+save_
+
+save_nef_molecular_system
+   _nef_molecular_system.sf_category   nef_molecular_system
+   _nef_molecular_system.sf_framecode  nef_molecular_system
+
+   loop_
+      _nef_sequence.index
+      _nef_sequence.chain_code
+      _nef_sequence.sequence_code
+      _nef_sequence.residue_name
+      _nef_sequence.linking
+      _nef_sequence.residue_variant
+      _nef_sequence.cis_peptide
+
+     1    A   2    MET   start    .   .
+     2    A   3    GLN   middle   .   .
+     3    A   4    ILE   middle   .   .
+     4    A   5    PHE   end      .   .
+
+
+   stop_
+
+save_
+
+save_nef_chemical_shift_list_nmrpipe
+   _nef_chemical_shift_list.sf_category   nef_chemical_shift_list
+   _nef_chemical_shift_list.sf_framecode  nef_chemical_shift_list_nmrpipe
+
+   loop_
+      _nef_chemical_shift.chain_code
+      _nef_chemical_shift.sequence_code
+      _nef_chemical_shift.residue_name
+      _nef_chemical_shift.atom_name
+      _nef_chemical_shift.value
+      _nef_chemical_shift.value_uncertainty
+      _nef_chemical_shift.element
+      _nef_chemical_shift.isotope_number
+
+     A   2    MET   CB    33.27    .   .   .
+     A   2    MET   CA    54.45    .   .   .
+     A   2    MET   C     170.54   .   .   .
+     A   2    MET   HA    4.23     .   .   .
+     A   3    GLN   HN    8.9      .   .   .
+     A   3    GLN   CA    55.08    .   .   .
+     A   3    GLN   CB    30.76    .   .   .
+     A   3    GLN   C     175.92   .   .   .
+     A   3    GLN   HA    5.249    .   .   .
+     A   3    GLN   N     123.22   .   .   .
+     A   4    ILE   HA    4.213    .   .   .
+     A   4    ILE   C     172.45   .   .   .
+     A   4    ILE   CA    59.57    .   .   .
+     A   4    ILE   N     115.34   .   .   .
+     A   4    ILE   HN    8.32     .   .   .
+     A   4    ILE   CB    42.21    .   .   .
+     A   5    PHE   HA    5.63     .   .   .
+     A   5    PHE   C     175.32   .   .   .
+     A   5    PHE   CB    41.48    .   .   .
+     A   5    PHE   CA    55.21    .   .   .
+     A   5    PHE   N     118.11   .   .   .
+     A   5    PHE   HN    8.61     .   .   .
+
+   stop_
+
+save_