Merge pull request #92 from kif/free_dnn

new application free_dnn

doc/source/changelog.rst

@@ -1,6 +1,12 @@
 Change-log
 ##########
 
+FreeSAS v2024.9.0 11/09/2021
+============================
+- Rg+Dmax inferance from a Dense Neural Network (free_dnn)
+- automatic packaging using github-actions
+- Supports Python 3.7-3.12, including numpy2
+
 FreeSAS v0.9.0 16/07/2021
 =========================
 - Command line tools upgrade

pyproject.toml

@@ -34,7 +34,8 @@ classifiers = [
 dependencies = [
  'numpy',
  'scipy',
- 'matplotlib'
+ 'matplotlib',
+ 'silx'
 ]
 [build-system]
 build-backend = 'mesonpy'
@@ -49,7 +50,8 @@ requires = [
  "numpy; platform_machine != 'ppc64le'",
  'pyproject-metadata>=0.5.0',
  'tomli>=1.0.0',
- 'scipy'
+ 'scipy',
+ 'silx'
 ]
 
 [project.urls]
@@ -67,6 +69,7 @@ cormapy = 'freesas.app.cormap:main'
 supycomb = 'freesas.app.supycomb:main'
 free_bift = 'freesas.app.bift:main'
 extract_ascii = 'freesas.app.extract_ascii:main'
+free_dnn = 'freesas.app.dnn:main'
 
 [project.gui-scripts]
 freesas = 'freesas.app.plot_sas:main'

src/freesas/app/dnn.py

@@ -0,0 +1,74 @@
+#!/usr/bin/python3
+# coding: utf-8
+#
+# Project: freesas
+# https://github.com/kif/freesas
+#
+# Copyright (C) 2020 European Synchrotron Radiation Facility, Grenoble, France
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+__author__ = ["Jérôme Kieffer", "Mayank Yadav"]
+__license__ = "MIT"
+__copyright__ = "2024, ESRF"
+__date__ = "11/09/2024"
+
+import sys
+import logging
+from freesas.sas_argparser import SASParser
+from freesas.fitting import run_dnn
+
+logging.basicConfig(level=logging.WARNING)
+logger = logging.getLogger("free_dnn")
+
+if sys.version_info < (3, 6):
+ logger.error("This code uses F-strings and requires Python 3.6+")
+
+
+def build_parser() -> SASParser:
+ """Build parser for input and return list of files.
+ :return: parser
+ """
+ description = (
+ "Assess the radius of gyration (Rg) and the diameter of the particle (Dmax) using a Dense Neural-Network"
+ " for a set of scattering curves"
+ )
+ epilog = """free_dnn is an alternative implementation of
+ `gnnom` (https://doi.org/10.1016/j.str.2022.03.011).
+ As this tool used a different training set, some results are likely to differ.
+ """
+ parser = SASParser(prog="free_gpa", description=description, epilog=epilog)
+ file_help_text = "dat files of the scattering curves"
+ parser.add_file_argument(help_text=file_help_text)
+ parser.add_output_filename_argument()
+ parser.add_output_data_format("native", "csv", "ssf", default="native")
+ parser.add_q_unit_argument()
+
+ return parser
+
+
+
+def main() -> None:
+ """Entry point for free_gpa app"""
+ parser = build_parser()
+ run_dnn(parser=parser, logger=logger)
+
+
+if __name__ == "__main__":
+ main()

src/freesas/app/meson.build

@@ -7,7 +7,8 @@ py.install_sources([
  'cormap.py',
  'extract_ascii.py',
  'plot_sas.py',
- 'supycomb.py'
+ 'supycomb.py',
+ 'dnn.py'
 ],
  pure: false, # Will be installed next to binaries
  subdir: 'freesas/app' # Folder relative to site-packages to install to

src/freesas/fitting.py

@@ -5,7 +5,7 @@
 __contact__ = "[email protected]"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "29/11/2023"
+__date__ = "11/09/2024"
 __status__ = "development"
 __docformat__ = "restructuredtext"
 
@@ -26,7 +26,7 @@
  load_scattering_data,
  convert_inverse_angstrom_to_nanometer,
 )
-from .sas_argparser import GuinierParser
+from .sas_argparser import GuinierParser, SASParser
 
 
 def set_logging_level(verbose_flag: int) -> None:
@@ -110,6 +110,28 @@ def get_guinier_header(
  else:
  return ""
 
+def get_dnn_header(
+ linesep: str, output_format: Optional[str] = None
+) -> str:
+ """Return appropriate header line for selected output format
+ :param output_format: output format from string parser
+ :param linesep: correct linesep for chosen destination
+ :return: a one-line string"""
+ # pylint: disable=R1705
+ if output_format == "csv":
+ return (
+ ",".join(
+ (
+ "File",
+ "Rg",
+ "Dmax",
+ )
+ )
+ + linesep
+ )
+ else:
+ return ""
+
 
 def rg_result_to_output_line(
  rg_result: RG_RESULT,
@@ -161,6 +183,44 @@ def rg_result_to_output_line(
  else:
  return f"{afile} {rg_result}{linesep}"
 
+def dnn_result_to_output_line(
+ dnn_result: tuple,
+ afile: Path,
+ linesep: str,
+ output_format: Optional[str] = None,
+) -> str:
+ """Return result line formatted according to selected output format
+ :param dnn_result: Result of an dnn inference, 2 tuple
+ :param afile: The name of the file that was processed
+ :param output_format: The chosen output format
+ :param linesep: correct linesep for chosen destination
+ :return: a one-line string including linesep"""
+ # pylint: disable=R1705
+ if output_format == "csv":
+ return (
+ ",".join(
+ [
+ f"{afile}",
+ f"{dnn_result[0]:6.4f}",
+ f"{dnn_result[1]:6.4f}",
+ ]
+ )
+ + linesep
+ )
+ elif output_format == "ssv":
+ return (
+ " ".join(
+ [
+ f"{dnn_result[0]:6.4f}",
+ f"{dnn_result[1]:6.4f}",
+ f"{afile}",
+ ]
+ )
+ + linesep
+ )
+ else:
+ return f"{afile} {dnn_result[0]} {dnn_result[1]}{linesep}"
+
 
 def run_guinier_fit(
  fit_function: Callable[[ndarray], RG_RESULT],
@@ -220,3 +280,61 @@ def run_guinier_fit(
  )
  output_destination.write(res)
  output_destination.flush()
+def run_dnn(
+ parser: SASParser,
+ logger: logging.Logger,
+) -> None:
+ """
+ reads in the data, infer the DNN and creates the result
+ :param parser: a function that returns the output of argparse.parse()
+ :param logger: a Logger
+ """
+ from .dnn import Rg_Dmax # heavy import
+
+ args = parser.parse_args()
+ set_logging_level(args.verbose)
+ files = collect_files(args.file)
+ logger.debug("%s input files", len(files))
+
+ with get_output_destination(args.output) as output_destination:
+ linesep = get_linesep(output_destination)
+
+ output_destination.write(
+ get_dnn_header(
+ linesep,
+ args.format,
+ )
+ )
+
+ for afile in files:
+ logger.info("Processing %s", afile)
+ try:
+ data = load_scattering_data(afile)
+ except OSError:
+ logger.error("Unable to read file %s", afile)
+ except ValueError:
+ logger.error("Unable to parse file %s", afile)
+ else:
+ if args.unit == "Å":
+ data = convert_inverse_angstrom_to_nanometer(data)
+ q, I = data.T[:2]
+ try:
+ dnn_result = Rg_Dmax(q, I)
+ except (
+ InsufficientDataError,
+ NoGuinierRegionError,
+ ValueError,
+ IndexError,
+ ) as err:
+ sys.stderr.write(
+ f"{afile}, {err.__class__.__name__}: {err} {os.linesep}"
+ )
+ else:
+ res = dnn_result_to_output_line(
+ dnn_result,
+ afile,
+ linesep,
+ args.format,
+ )
+ output_destination.write(res)
+ output_destination.flush()

version.py

@@ -52,7 +52,7 @@
 __authors__ = ["Jérôme Kieffer"]
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "04/12/2023"
+__date__ = "11/09/2024"
 __status__ = "production"
 __docformat__ = 'restructuredtext'
 __all__ = ["date", "version_info", "strictversion", "hexversion", "debianversion",
@@ -68,11 +68,11 @@
  "alpha": "a",
  "beta": "b",
  "candidate": "rc"}
-MAJOR = 0
+MAJOR = 2024
 MINOR = 9
-MICRO = 9
+MICRO = 0
 RELEV = "dev" # <16
-SERIAL = 1 # <16
+SERIAL = 0 # <16
 
 date = __date__