Add ML finding and other new developments

package/CMakeLists.txt

@@ -50,7 +50,7 @@ FetchContent_Declare(
 # Download Gemmi
 FetchContent_Declare(
         gemmi-dependencies
-        URL http://www.ysbl.york.ac.uk/jsd523/gemmi-0.6.5.tar.gz
+        URL https://github.com/Dialpuri/gemmi-bundles/raw/refs/heads/main/gemmi-0.7.3.tar.gz
 )
 
 FetchContent_MakeAvailable(clipper-dependencies mmdb2-dependencies fftw-dependencies ccp4-dependencies gemmi-dependencies)
@@ -101,8 +101,10 @@ add_library(
         ${WRK_DIR}/src/cpp/sails-telemetry.cpp
         ${WRK_DIR}/src/cpp/sails-solvent.cpp
         ${WRK_DIR}/src/cpp/sails-wurcs.cpp
-
+        ${WRK_DIR}/src/cpp/sails-predictions.cpp
         ${WRK_DIR}/src/cpp/sails-morph.cpp
+        ${WRK_DIR}/src/cpp/sails-score.cpp
+
         # Density
         ${WRK_DIR}/src/cpp/density/sails-density.cpp
         ${WRK_DIR}/src/cpp/density/sails-xtal-density.cpp

package/_pyproject.toml

@@ -18,7 +18,10 @@ dependencies=[
     'tqdm',
     'gemmi',
     'numpy<2.0.0',
-    'typing-extensions'
+    'typing-extensions',
+    'onnxruntime-gpu; platform_system != "Darwin"',
+    'onnxruntime; platform_system == "Darwin"',
+    'requests'
 ]
 
 [project.urls]
@@ -32,6 +35,10 @@ sails-find = "sails.find:run"
 sails-test = "sails.test:run"
 sails-wurcs = "sails.wurcs:run"
 sails-morph = "sails.morph:run"
+sails-predict = "sails.prediction.predict:run"
+sails-install = "sails.install:run"
+sails-clean = "sails.clean:run"
+sails-validate = "sails.validate:run"
 
 [tool.scikit-build]
 # Protect the configuration against future changes in scikit-build-core

package/gemmi/CMakeLists.txt

@@ -22,24 +22,31 @@ add_library(gemmi_cpp STATIC
         ${gemmi_src}/src/align.cpp
         ${gemmi_src}/src/assembly.cpp
         ${gemmi_src}/src/calculate.cpp
+        ${gemmi_src}/src/ccp4.cpp
         ${gemmi_src}/src/crd.cpp
         ${gemmi_src}/src/ddl.cpp
+        ${gemmi_src}/src/dssp.cpp
         ${gemmi_src}/src/eig3.cpp
+        ${gemmi_src}/src/fprime.cpp
         ${gemmi_src}/src/gz.cpp
         ${gemmi_src}/src/intensit.cpp
         ${gemmi_src}/src/json.cpp
         ${gemmi_src}/src/mmcif.cpp
         ${gemmi_src}/src/mmread_gz.cpp
+        ${gemmi_src}/src/monlib.cpp
         ${gemmi_src}/src/mtz.cpp
         ${gemmi_src}/src/mtz2cif.cpp
         ${gemmi_src}/src/polyheur.cpp
+        ${gemmi_src}/src/pdb.cpp
         ${gemmi_src}/src/read_cif.cpp
         ${gemmi_src}/src/resinfo.cpp
         ${gemmi_src}/src/riding_h.cpp
+        ${gemmi_src}/src/select.cpp
         ${gemmi_src}/src/sprintf.cpp
+        ${gemmi_src}/src/symmetry.cpp
+        ${gemmi_src}/src/to_json.cpp
         ${gemmi_src}/src/to_mmcif.cpp
         ${gemmi_src}/src/to_pdb.cpp
-        ${gemmi_src}/src/monlib.cpp
         ${gemmi_src}/src/topo.cpp
         ${gemmi_src}/src/xds_ascii.cpp
 )
@@ -63,7 +70,6 @@ set(gemmi_HEADERS
         ${gemmi_src}/include/gemmi/ccp4.hpp
         ${gemmi_src}/include/gemmi/cellred.hpp
         ${gemmi_src}/include/gemmi/chemcomp.hpp
-        ${gemmi_src}/include/gemmi/chemcomp_xyz.hpp
         ${gemmi_src}/include/gemmi/cif.hpp
         ${gemmi_src}/include/gemmi/cif2mtz.hpp
         ${gemmi_src}/include/gemmi/cifdoc.hpp
@@ -72,6 +78,7 @@ set(gemmi_HEADERS
         ${gemmi_src}/include/gemmi/ddl.hpp
         ${gemmi_src}/include/gemmi/dencalc.hpp
         ${gemmi_src}/include/gemmi/dirwalk.hpp
+        ${gemmi_src}/include/gemmi/dssp.hpp
         ${gemmi_src}/include/gemmi/ecalc.hpp
         ${gemmi_src}/include/gemmi/eig3.hpp
         ${gemmi_src}/include/gemmi/elem.hpp
@@ -86,20 +93,21 @@ set(gemmi_HEADERS
         ${gemmi_src}/include/gemmi/grid.hpp
         ${gemmi_src}/include/gemmi/gz.hpp
         ${gemmi_src}/include/gemmi/input.hpp
+        ${gemmi_src}/include/gemmi/intensit.hpp
         ${gemmi_src}/include/gemmi/interop.hpp
         ${gemmi_src}/include/gemmi/it92.hpp
         ${gemmi_src}/include/gemmi/iterator.hpp
         ${gemmi_src}/include/gemmi/json.hpp
         ${gemmi_src}/include/gemmi/levmar.hpp
         ${gemmi_src}/include/gemmi/linkhunt.hpp
+        ${gemmi_src}/include/gemmi/logger.hpp
         ${gemmi_src}/include/gemmi/math.hpp
-#        ${gemmi_src}/include/gemmi/merge.hpp
         ${gemmi_src}/include/gemmi/metadata.hpp
-        ${gemmi_src}/include/gemmi/mmcif.hpp
         ${gemmi_src}/include/gemmi/mmcif_impl.hpp
+        ${gemmi_src}/include/gemmi/mmcif.hpp
         ${gemmi_src}/include/gemmi/mmdb.hpp
-        ${gemmi_src}/include/gemmi/mmread.hpp
         ${gemmi_src}/include/gemmi/mmread_gz.hpp
+        ${gemmi_src}/include/gemmi/mmread.hpp
         ${gemmi_src}/include/gemmi/model.hpp
         ${gemmi_src}/include/gemmi/modify.hpp
         ${gemmi_src}/include/gemmi/monlib.hpp
@@ -108,24 +116,23 @@ set(gemmi_HEADERS
         ${gemmi_src}/include/gemmi/neighbor.hpp
         ${gemmi_src}/include/gemmi/neutron92.hpp
         ${gemmi_src}/include/gemmi/numb.hpp
-        ${gemmi_src}/include/gemmi/pdb.hpp
         ${gemmi_src}/include/gemmi/pdb_id.hpp
+        ${gemmi_src}/include/gemmi/pdb.hpp
         ${gemmi_src}/include/gemmi/pirfasta.hpp
         ${gemmi_src}/include/gemmi/polyheur.hpp
         ${gemmi_src}/include/gemmi/qcp.hpp
         ${gemmi_src}/include/gemmi/read_cif.hpp
-        ${gemmi_src}/include/gemmi/read_map.hpp
         ${gemmi_src}/include/gemmi/recgrid.hpp
         ${gemmi_src}/include/gemmi/reciproc.hpp
         ${gemmi_src}/include/gemmi/refln.hpp
-        ${gemmi_src}/include/gemmi/remarks.hpp
         ${gemmi_src}/include/gemmi/resinfo.hpp
         ${gemmi_src}/include/gemmi/riding_h.hpp
         ${gemmi_src}/include/gemmi/scaling.hpp
         ${gemmi_src}/include/gemmi/select.hpp
         ${gemmi_src}/include/gemmi/seqalign.hpp
         ${gemmi_src}/include/gemmi/seqid.hpp
         ${gemmi_src}/include/gemmi/seqtools.hpp
+        ${gemmi_src}/include/gemmi/serialize.hpp
         ${gemmi_src}/include/gemmi/sfcalc.hpp
         ${gemmi_src}/include/gemmi/small.hpp
         ${gemmi_src}/include/gemmi/smcif.hpp
@@ -146,14 +153,14 @@ set(gemmi_HEADERS
         ${gemmi_src}/include/gemmi/util.hpp
         ${gemmi_src}/include/gemmi/version.hpp
         ${gemmi_src}/include/gemmi/xds_ascii.hpp
-
+        ${gemmi_src}/include/gemmi/xds2mtz.hpp
 
 )
 
 set(gemmi_third_party-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/fast_float.h
         ${gemmi_src}/include/gemmi/third_party/pocketfft_hdronly.h
-#        ${gemmi_src}/include/gemmi/third_party/sajson.h
+        #        ${gemmi_src}/include/gemmi/third_party/sajson.h
         ${gemmi_src}/include/gemmi/third_party/tinydir.h
 )
 
@@ -198,7 +205,7 @@ set(gemmi_third_party_tao_pegtl_analysis-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/generic.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/grammar_info.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/insert_guard.hpp
-        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/insert_rules.hpp
+        #        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/insert_rules.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/rule_info.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/analysis/rule_type.hpp
 )
@@ -217,7 +224,7 @@ set(gemmi_third_party_tao_pegtl_internal-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bof.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bol.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bump_help.hpp
-        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bump_impl.hpp
+        #        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bump_impl.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/bytes.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/control.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/cr_crlf_eol.hpp
@@ -240,7 +247,7 @@ set(gemmi_third_party_tao_pegtl_internal-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/eof.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/eol.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/eolf.hpp
-        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/file_mapper.hpp
+        #        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/file_mapper.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/file_opener.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/file_reader.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/has_apply.hpp
@@ -262,7 +269,7 @@ set(gemmi_third_party_tao_pegtl_internal-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/list_tail.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/list_tail_pad.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/marker.hpp
-        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/minus.hpp
+        #        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/minus.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/must.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/not_at.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/one.hpp
@@ -284,7 +291,7 @@ set(gemmi_third_party_tao_pegtl_internal-headers_HEADERS
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/rep_opt.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/require.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/result_on_found.hpp
-        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/rule_conjunction.hpp
+        #        ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/rule_conjunction.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/rules.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/seq.hpp
         ${gemmi_src}/include/gemmi/third_party/tao/pegtl/internal/skip_control.hpp

package/pyproject.toml

@@ -18,11 +18,15 @@ dependencies=[
     'tqdm',
     'gemmi',
     'numpy<2.0.0',
-    'typing-extensions'
+    'typing-extensions',
+    'onnxruntime-gpu; platform_system != "Darwin"',
+    'onnxruntime; platform_system == "Darwin"',
+    'requests',
+    'nanobind==2.4.0'
 ]
 
 [tool.setuptools]
-packages = ["sails"]
+packages = ["sails", "sails.prediction"]
 package-dir = {"" = "src"}
 include-package-data = true
 
@@ -40,6 +44,10 @@ sails-find = "sails.find:run"
 sails-test = "sails.test:run"
 sails-wurcs = "sails.wurcs:run"
 sails-morph = "sails.morph:run"
+sails-predict = "sails.prediction.predict:run"
+sails-install = "sails.install:run"
+sails-clean = "sails.clean:run"
+sails-validate = "sails.validate:run"
 
 [tool.scikit-build]
 # Protect the configuration against future changes in scikit-build-core

package/scripts/compare_structures.py

@@ -2,6 +2,7 @@
 import gemmi
 import json
 import numpy as np
+from pprint import pprint
 
 
 def load_data_file(filename):
@@ -20,11 +21,11 @@ def format_residue(chain: gemmi.Chain, residue: gemmi.Residue):
 
 
 def main(args):
-    data = load_data_file("package/data/data.json")
+    data = load_data_file("package/src/sails/data/data.json")
     structure = gemmi.read_structure(args.model)
     reference = gemmi.read_structure(args.reference)
 
-    ns = gemmi.NeighborSearch(structure, max_radius=1).populate()
+    ns = gemmi.NeighborSearch(structure, max_radius=1.5).populate()
 
     output = {}
 
@@ -84,6 +85,8 @@ def main(args):
     percentage_modelled = 100 * modelled / total_sugars
     print(f"Percentage Modelled {percentage_modelled:.2f}")
 
+    pprint(output)
+
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()