1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
|
// Copyright 2017 Global Phasing Ltd.
#include "gemmi/version.hpp" // for GEMMI_VERSION
#include "gemmi/math.hpp" // for hc
#include "gemmi/dirwalk.hpp" // for CifWalk, CoorFileWalk
#include "gemmi/pdb_id.hpp" // for expand_if_pdb_code
#include "gemmi/bessel.hpp" // for bessel_i1_over_i0
#include "gemmi/pirfasta.hpp" // for read_pir_or_fasta
#include "gemmi/seqtools.hpp" // for calculate_sequence_weight
#include "gemmi/stats.hpp" // for Correlation
#include "tao/pegtl/parse_error.hpp" // for parse_error
#include "common.h"
#include "array.h"
#include <nanobind/make_iterator.h>
#include <nanobind/stl/string.h>
#include <nanobind/stl/vector.h> // for calculate_sequence_weight
namespace {
// returns max value in the array bin indices (which are of type int)
template<typename T> size_t get_max_bin(const T& bins) {
int max_bin = 0;
for (size_t i = 0; i < bins.shape(0); ++i) {
if (bins(i) < 0)
throw nb::value_error("bins argument must have no negative elements");
max_bin = std::max(max_bin, bins(i));
}
if (max_bin > 1000000)
throw nb::value_error("bin numbers must be smaller than million");
return (size_t) max_bin;
}
struct VectorizeFunc {
typedef double (*Func)(double);
Func func;
auto operator()(const cpu_array<double>& x) const {
auto x_ = x.view();
size_t len = x_.shape(0);
auto ret = make_numpy_array<double>({len});
double* retp = ret.data();
for (size_t i = 0; i < len; ++i)
retp[i] = func(x_(i));
return ret;
}
};
void add_misc(nb::module_& m) {
nb::class_<gemmi::CifWalk>(m, "CifWalk")
.def(nb::init<const char*, char>(), nb::arg("path"), nb::arg("try_pdbid")='\0')
.def("__iter__", [](gemmi::CifWalk& self) {
return nb::make_iterator(nb::type<gemmi::CifWalk>(), "iterator", self);
}, nb::keep_alive<0, 1>());
nb::class_<gemmi::CoorFileWalk>(m, "CoorFileWalk")
.def(nb::init<const char*, char>(), nb::arg("path"), nb::arg("try_pdbid")='\0')
.def("__iter__", [](gemmi::CoorFileWalk& self) {
return nb::make_iterator(nb::type<gemmi::CoorFileWalk>(), "iterator", self);
}, nb::keep_alive<0, 1>());
m.def("is_pdb_code", &gemmi::is_pdb_code);
m.def("path_in_pdb_dir", &gemmi::path_in_pdb_dir);
m.def("expand_pdb_code_to_path", &gemmi::expand_pdb_code_to_path,
nb::arg("code"), nb::arg("filetype"), nb::arg("throw_if_unset")=false);
m.def("expand_if_pdb_code", &gemmi::expand_if_pdb_code,
nb::arg("code"), nb::arg("filetype")='M');
m.attr("hc") = nb::float_(gemmi::hc());
m.def("bessel_i1_over_i0", VectorizeFunc{gemmi::bessel_i1_over_i0});
m.def("log_bessel_i0", VectorizeFunc{gemmi::log_bessel_i0});
m.def("log_cosh", VectorizeFunc{gemmi::log_cosh});
// pirfasta.hpp
nb::class_<gemmi::FastaSeq>(m, "FastaSeq")
.def_ro("header", &gemmi::FastaSeq::header)
.def_ro("seq", &gemmi::FastaSeq::seq)
;
m.def("read_pir_or_fasta", &gemmi::read_pir_or_fasta);
// seqtools.hpp
m.def("calculate_sequence_weight", &gemmi::calculate_sequence_weight,
nb::arg("sequence"), nb::arg("unknown")=0.);
m.def("one_letter_code", &gemmi::one_letter_code);
m.def("pdbx_one_letter_code", &gemmi::pdbx_one_letter_code);
m.def("sequence_kind", &gemmi::sequence_kind);
// stats.hpp
nb::class_<gemmi::Correlation>(m, "Correlation")
.def_ro("n", &gemmi::Correlation::n)
.def("coefficient", &gemmi::Correlation::coefficient)
.def("mean_ratio", &gemmi::Correlation::mean_ratio)
;
// utilities inspired by numpy.bincount()
m.def("binmean", [](const cpu_array<int>& bins, const cpu_array<double>& values) {
auto bins_ = bins.view();
auto values_ = values.view();
auto len = bins_.shape(0);
if (len != values_.shape(0))
throw std::domain_error("arrays have different lengths");
size_t ret_size = get_max_bin(bins_) + 1;
auto ret = make_numpy_array<double>({ret_size});
double* retp = ret.data();
for (size_t i = 0; i != ret_size; ++i)
retp[i] = 0.;
std::vector<int> counts(ret_size);
for (size_t i = 0; i != len; ++i)
if (!std::isnan(values_(i))) {
int n = bins_(i);
counts[n]++;
retp[n] += values_(i);
}
for (size_t i = 0; i != ret_size; ++i)
retp[i] /= counts[i];
return ret;
}, nb::arg("nbins"), nb::arg("values"));
m.def("binrfactor", [](const cpu_array<int>& bins, const cpu_array<double>& obs,
const cpu_array<double>& calc, bool riso) {
auto bins_ = bins.view();
auto obs_ = obs.view();
auto calc_ = calc.view();
auto len = bins_.shape(0);
if (len != obs_.shape(0) || len != calc_.shape(0))
throw std::domain_error("arrays have different lengths");
size_t ret_size = get_max_bin(bins_) + 1;
auto ret = make_numpy_array<double>({ret_size});
double* retp = ret.data();
for (size_t i = 0; i != ret_size; ++i)
retp[i] = 0.;
std::vector<double> sum_fobs(ret_size);
for (size_t i = 0; i != len; ++i)
if (!std::isnan(obs_(i)) && !std::isnan(calc_(i))) {
int n = bins_(i);
retp[n] += std::fabs(obs_(i) - calc_(i));
sum_fobs[n] += riso ? (obs_(i) + calc_(i)) : obs_(i);
}
for (size_t i = 0; i != ret_size; ++i)
retp[i] /= (riso ? 0.5 * sum_fobs[i] : sum_fobs[i]);
return ret;
}, nb::arg("nbins"), nb::arg("obs"), nb::arg("calc"), nb::arg("riso")=false);
m.def("bincorr", [](const cpu_array<int>& bins, const cpu_array<double>& obs,
const cpu_array<double>& calc) {
auto bins_ = bins.view();
auto obs_ = obs.view();
auto calc_ = calc.view();
auto len = bins_.shape(0);
if (len != obs_.shape(0) || len != calc_.shape(0))
throw std::domain_error("arrays have different lengths");
size_t ret_size = get_max_bin(bins_) + 1;
std::vector<gemmi::Correlation> cor(ret_size);
for (size_t i = 0; i != len; ++i)
if (!std::isnan(obs_(i)) && !std::isnan(calc_(i)))
cor[bins_(i)].add_point(obs_(i), calc_(i));
return cor;
}, nb::arg("nbins"), nb::arg("obs"), nb::arg("calc"));
}
} // anonymous namespace
NB_MODULE(gemmi_ext, m_) {
// unusual setup: importing gemmi_ext adds classes and functions to gemmi
(void) m_;
nb::module_ m = nb::module_::import_("gemmi");
m.doc() = "Python bindings to GEMMI - a library used in macromolecular\n"
"crystallography and related fields";
m.attr("__version__") = GEMMI_VERSION;
#ifdef NDEBUG
nb::set_leak_warnings(false);
#endif
nb::register_exception_translator([](const std::exception_ptr& p, void*) {
try {
if (p)
std::rethrow_exception(p);
} catch (const std::system_error &e) {
const int errornum = e.code().value();
PyErr_SetObject(PyExc_IOError, nb::make_tuple(errornum, e.what()).ptr());
} catch (const tao::pegtl::parse_error &e) {
PyErr_SetString(PyExc_ValueError, e.what());
}
});
nb::module_ mcif = m.def_submodule("cif", "CIF file format");
add_cif(mcif);
add_symmetry(m);
add_unitcell(m);
add_elem(m);
add_xds(m);
add_meta(m);
add_mol(m);
add_small(m);
add_misc(m);
add_grid(m);
add_recgrid(m);
add_ccp4(m);
add_sf(m);
add_cif_read(mcif);
add_mtz(m);
add_hkl(m);
add_chemcomp(m);
add_monlib(m);
add_topo(m);
add_alignment(m);
add_search(m);
add_read_structure(m);
add_scaling(m);
m.def("set_leak_warnings", nb::set_leak_warnings);
}
|
