svmfp related

contrib/bin/calibrate_svmfp_client.rb

@@ -0,0 +1,91 @@
+#!/usr/bin/env ruby
+#!/usr/bin/env ruby
+
+require 'fileutils'
+
+# Client script for model_calibrate. Builds and scores svmfp models
+# using svmfp_make and svmfp_evaluate.
+
+require_relative 'lib/iwcmdline'
+
+def usage(rc)
+  exit(rc)
+end
+
+def main
+  cl = IWCmdline.new("-v-gfp=close-TRSMI=sfile-TESMI=sfile-TRactivity=sfile-TEactivity=sfile-PRED=s-STATS=s-TMPDIR=s-uid=s")
+
+  unless cl.option_present('gfp')
+    $stderr << "Must specify fingerprints via the -gfp option\n"
+    usage(1)
+  end
+
+  unless cl.option_present('TRactivity')
+    $stderr << "Must specify training set activity file via the -TRactivity option\n"
+    usage(1)
+  end
+
+  unless cl.option_present('TEactivity')
+    $stderr << "Must specify testing set activity file via the -TEactivity option\n"
+    usage(1)
+  end
+
+  unless cl.option_present('PRED')
+    $stderr << "Must specify predicted values file via the -PRED option\n"
+    usage(1)
+  end
+
+  unless cl.option_present('STATS')
+    $stderr << "Must specify statistics file via the -STATS option\n"
+    usage(1)
+  end
+
+  if cl.option_present('TMPDIR')
+    tmpdir = cl.value('TMPDIR')
+  elsif cl.option_present('uid')
+    uid = cl.value('uid')
+    tmpdir = "/tmp/calibrate_#{uid}"
+  else
+    $stderr << "Must specify either -TMPDIR or unique identifier via the -uid option\n"
+    usage(1)
+  end
+
+  trsmi = cl.value('TRSMI')
+  tesmi = cl.value('TESMI')
+
+  Dir.mkdir(tmpdir) unless File.directory?(tmpdir)
+
+  if cl.unrecognised_options_encountered
+    $stderr << "Unrecognised options encountered\n"
+    usage(1)
+  end
+
+  verbose = cl.option_present('v')
+
+  lillymol_home = ENV['LILLYMOL_HOME']
+  svmfp_make = "#{lillymol_home}/contrib/bin/svmfp/svmfp_make.sh"
+  svmfp_evaluate = "#{lillymol_home}/contrib/bin/svmfp/svmfp_evaluate.sh"
+
+  gfp = cl.value('gfp')
+  train_activity = cl.value('TRactivity')
+  test_activity = cl.value('TEactivity')
+
+  predicted = cl.value('PRED')
+
+  results = cl.value('STATS')
+
+  mdir = File.join(tmpdir, 'MODEL')
+
+  cmd = "#{svmfp_make} --mdir #{mdir} -gfp #{gfp} -gfp -A #{train_activity} #{trsmi}"
+  system(cmd)
+
+  cmd = "#{svmfp_evaluate} -mdir #{mdir} #{tesmi} > #{predicted}"
+  system(cmd)
+
+  cmd = "iwstats -w -Y allequals -E #{test_activity} -p 2 #{predicted} > #{results}"
+  system(cmd)
+
+  FileUtils.rm_rf(tmpdir)
+end
+
+main

contrib/bin/calibrate_svmfp_client.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+set -x
+ruby_script="${0%%.sh}.rb"
+exec ruby ${ruby_script} "$@"

contrib/bin/calibrate_xgbd_client.rb

@@ -86,9 +86,6 @@ def main
   system(cmd)
 
   FileUtils.rm_rf(tmpdir)
-
-  File.unlink(tmptrain)
-  File.unlink(tmptest)
 end
 
 main

contrib/bin/calibrate_xgbd_client.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+ruby_script="${0%%.sh}.rb"
+exec ruby ${ruby_script} "$@"

contrib/bin/gfp_make.pl

@@ -109,11 +109,18 @@
 
 my $psa = 0;
 my $psa_bit_replicates = 0;
+
 my $abr = 0;
 my $abr_bit_replicates = 0;
 my $abp = 0;
 my $abp_bit_replicates = 0;
 
+my $alogp = 0;
+my $alogp_bit_replicates = 0;
+
+my $xlogp = 0;
+my $xlogp_bit_replicates = 0;
+
 my $temperature = 0;
 my $natoms = 0;
 my $nrings = 0;
@@ -258,6 +265,8 @@ sub usage
   print STDERR " -MVD           FP Marvin logd 7.4\n" if ($expert);
   print STDERR " -PSA           FP Novartis Polar Surface Area\n" if ($expert);
   print STDERR " -ABR           FP Abraham fingerprint\n" if ($expert);
+  print STDERR " -ALOGP         FP alogp logp\n";
+  print STDERR " -XLOGP         FP xlogp logp\n";
   print STDERR " -ABP           FP Abraham and Platts fingerprint\n" if ($expert);
   print STDERR " -INS ... -INS  FP insert arbitrary pipelined commands, F:... to read from file\n" if ($expert);
   print STDERR " -D2F ... -D2F  FP use descriptors_to_fingerprint to insert bucketised descriptors (very flexible)\n" if ($expert);
@@ -1364,6 +1373,28 @@ sub ust_needed_for_atype
     $psa_bit_replicates = $1;
     $fingerprints_specified++;
   }
+  elsif ($opt eq "-ALOGP")
+  {
+    $alogp = 1;
+    $fingerprints_specified++;
+  }
+  elsif ($opt =~ /^-ALOGP(\d+)/)
+  {
+    $alogp = 1;
+    $alogp_bit_replicates = $1;
+    $fingerprints_specified++;
+  }
+  elsif ($opt eq "-XLOGP")
+  {
+    $xlogp = 1;
+    $fingerprints_specified++;
+  }
+  elsif ($opt =~ /^-XLOGP(\d+)/)
+  {
+    $xlogp = 1;
+    $xlogp_bit_replicates = $1;
+    $fingerprints_specified++;
+  }
   elsif ($opt eq "-W")
   {
     $w++;
@@ -2757,6 +2788,68 @@ sub flush_args {
   $psa_cmd_pipe = "${psa_cmd} -f -";
 }
 
+my $alogp_cmd_first;
+my $alogp_cmd_pipe;
+
+if ($alogp)
+{
+  my $alogp_exe = find_executable("alogp");
+  my $alogp_cmd = "${alogp_exe}";
+
+  if ($alogp_bit_replicates > 0)
+  {
+    $alogp_cmd .= " -p ${alogp_bit_replicates} -J NCALOGP${alogp_bit_replicates}"
+  }
+  else
+  {
+    $alogp_cmd .= " -J NCALOGP"
+  }
+
+  if ($work_as_filter)
+  {
+    $alogp_cmd_first = "${alogp_cmd} ${dash_g} ${aromatic_smiles} FILE";
+  }
+  elsif ($work_as_tdt_filter)
+  {
+    $alogp_cmd_first = "${alogp_cmd} ${dash_g} ${aromatic_smiles} -f -";
+  }
+  else
+  {
+    $alogp_cmd_first = "${alogp_cmd} ${dash_g} ${aromatic_smiles} FILE";
+  }
+
+  $alogp_cmd_pipe = "${alogp_cmd} -f -";
+}
+
+my $xlogp_cmd_first;
+my $xlogp_cmd_pipe;
+
+if ($xlogp)
+{
+  my $xlogp_exe = find_executable("xlogp");
+  my $xlogp_cmd = "${xlogp_exe}";
+
+  if ($xlogp_bit_replicates > 0)
+  {
+    $xlogp_cmd .= " -p ${xlogp_bit_replicates} -J NCXLOGP${xlogp_bit_replicates}"
+  }
+
+  if ($work_as_filter)
+  {
+    $xlogp_cmd_first = "${xlogp_cmd} ${dash_g} ${aromatic_smiles} FILE";
+  }
+  elsif ($work_as_tdt_filter)
+  {
+    $xlogp_cmd_first = "${xlogp_cmd} ${dash_g} ${aromatic_smiles} -f -";
+  }
+  else
+  {
+    $xlogp_cmd_first = "${xlogp_cmd} ${dash_g} ${aromatic_smiles} FILE";
+  }
+
+  $xlogp_cmd_pipe = "${xlogp_cmd} -f -";
+}
+
 my $abr_cmd_first;
 my $abr_cmd_pipe;
 
@@ -3249,6 +3342,32 @@ sub flush_args {
 
     $psa = 0;
   }
+  elsif ($alogp)
+  {
+    if ($first)
+    {
+      $cmd = $alogp_cmd_first;
+    }
+    else
+    {
+      $cmd .= "| $alogp_cmd_pipe";
+    }
+
+    $alogp = 0;
+  }
+  elsif ($xlogp)
+  {
+    if ($first)
+    {
+      $cmd = $xlogp_cmd_first;
+    }
+    else
+    {
+      $cmd .= "| $xlogp_cmd_pipe";
+    }
+
+    $xlogp = 0;
+  }
   elsif ($marvin)
   {
     if ($first)

contrib/bin/gfp_make.rb

@@ -22,7 +22,7 @@ def usage(config_dirs, retcode)
   $stderr << " will pass '-a 1 -B foo' to the FOO fingerprint generator\n"
   $stderr << " Most fingerprints are sparse, non colliding form. To generate a fixed width version\n";
   $stderr << " append :fixed to the fingerprint name, EC3:AQT:fixed generates a 2048 bit EC fingerprint\n"
-  $stderr << " Some fingerprints that are by default fixed, will response to :sparse as a modifier\n"
+  $stderr << " Some fingerprints that are by default fixed, will responsd to :sparse as a modifier\n"
   $stderr << "\n"
   $stderr << " -all ... -all    common options passed to all programmes\n"
   $stderr << " -v               verbose output\n"
@@ -138,9 +138,9 @@ def usage(config_dirs, retcode)
 fps = config_fingerprints(config_dirs, verbose)
 
 # List fingerprints recoginsed
-# fps.each do |k, v|
-#   $stderr << "#{k} #{v}\n"
-# end
+ fps.each do |k, v|
+   $stderr << "#{k} #{v}\n"
+ end
 
 # For each fp_option, a mapping to an object that knows how to
 # generate that command line component.

contrib/bin/model_calibrate.rb

@@ -20,6 +20,9 @@ def read_fingerprints(fnames)
   result = []
   fnames.each do |fname|
     File.readlines(fname).each do |line|
+      line.chomp!
+      next if line.empty?
+      next if line[0] == '#'
       result << line.chomp
     end
   end
@@ -148,9 +151,10 @@ def write_command_file(splits, descriptor_files, fingerprints, predicted_stem, s
     fptxt = fp.gsub(' ', "")
     splits.each_with_index do |split, ndx|
       file << "calibrate_svmfp_client.sh -gfp #{fp} -gfp " +
-              "-TRsmi #{split.train_smi} -TRactivity #{split.train_activity} " +
-              "-TEsmi #{split.test_smi} -TEactivity #{split.test_activity} " +
-              "-PRED #{predicted_stem}.#{fptxt}.#{ndx} -STATS #{stats_stem}.#{fptxt}.#{ndx}" +
+              "-TRSMI #{split.train_smi} -TRactivity #{split.train_activity} " +
+              "-TESMI #{split.test_smi} -TEactivity #{split.test_activity} " +
+              "-PRED #{predicted_stem}.#{fptxt}.#{ndx} -STATS #{stats_stem}.#{fptxt}.#{ndx} " +
+              "-uid SVMFP#{fptxt}" +
               "\n"
       $stderr << "Wrote #{fp} split #{ndx}\n"
     end

contrib/bin/svmfp/lib/fp_config_files.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+# Part of gfp_make
+
+# Report to stderr the recognized fingerprints in `fps`.
+def report_recognized_fingerprints(fps)
+  $stderr << "#{fps.length} fingerprints recognized\n"
+  fps.each do |k, v|
+    $stderr << "#{k} #{v} #{v.description}\n"
+  end
+end
+
+# Return a map of fingerprint names to object that can process
+# that kind of fingerprint.
+# By convention, the name of the class is the uppercase of
+# the config file name.
+# config_dirs is an array of directories in which to look.
+# If multiple items are found, the last will be silently used.
+
+def config_fingerprints(config_dirs, verbose) # rubocop:disable Metrics/AbcSize, Metrics/CyclomaticComplexity, Metrics/MethodLength, Metrics/PerceivedComplexity
+  fps = {} # To be returned.
+  config_dirs.each do |dir|
+    $stderr << "Fetching configs from #{dir}\n" if verbose.positive?
+    Dir.entries(dir).each do |fname|
+      next if /^\./.match(fname)
+      next if fname == 'fp_common.rb'
+      next if fname == 'lib'
+
+      m = /^(\S+)\.rb/.match(fname)
+      unless m
+        $stderr << "No match file name #{fname}, ignored\n"
+        next
+      end
+      require("#{dir}/#{fname}")
+      class_name = m[1].upcase # By convention.
+      fps[class_name] = eval("#{class_name}.new", binding, __FILE__, __LINE__)  # rubocop:disable Security/Eval
+    end
+  end
+
+  report_recognized_fingerprints(fps) if verbose.positive?
+
+  fps
+end