add_entry_v2 | searchcode

/scripts/add_entry_v2

https://bitbucket.org/JnWng/viperdb-analysis
PHP | 631 lines | 511 code | 65 blank | 55 comment | 81 complexity | a0d4b3dcc0d7deb8ad7417862b486955 MD5 | raw file

#!/usr/bin/php -q
<?php
include("viper_connect.php");
include("fetch_info.php");
include("cli_compatibility.php");

function field_input($name, $default) {
//  if($default==NULL) $default="NA";
  print "$name [$default]: ";
  $value = trim(fgets(STDIN));
  if(!$value) $value = $default;
  return $value;
}

$admin_base_dir = "/usr/people/viperdb/www/html/Admin";

system("clear");
print "\nThis is add_entry. Always press enter for [default] value.\n\n";

#$database = field_input("Database to use:", "viperdb");
#$database = trim($database);
$server= "viperdb";
$result = @mysql_pconnect($server, "viperdbadmin", "ccmv1m4x");
if(!$result) {
  print "Failed to connect to database $database.\n";
  exit();
}
$database = "viperdb";
if(!mysql_select_db($database)) {
  print "Failed to select database $database.\n";
  exit();
}

print "Enter entry (PDB) id for new entry: ";
$vdb = trim(fgets(STDIN));
$vdb = strtolower($vdb);
$query = "select entry_key from MMS_ENTRY where id = '$vdb'";
$result = mysql_query($query);
$entry_exists = mysql_num_rows($result);
#if it exists delete?
if($entry_exists){
  print "  WARNING: Entry $vdb already is in the db.\n";
  print "  The entry WILL BE DELETED if you press 'Enter'. Ctrl-C to quit.\n";
  fgets(STDIN);
  $delete_entry = field_input("    Are you sure you want to delete entry $vdb (Y|N)", "N");
  if($delete_entry=="Y"){
            $row = mysql_fetch_array($result);
            $entry_key = $row[0];
            mysql_free_result($result);
            $cmd="./delete_entry.pl -d $database -e $entry_key";
            system($cmd);
  }
  else{ print "\nThank you for using add_entry\n";exit(); }
}
#if it doesn't exist create
else{
  print "  Entry $vdb is new\n";
}
  $cmd="./pdbase_batch $vdb";
  print "\n  Creating it in db: '$cmd'\n";
  
   $output = array();
   $return_var = 0;
   $failed=0;
   exec($cmd, $output, $return_var);
   //echo "\n  return val: $return_var\n";   /* 126 */
   foreach($output as $o){
     echo "$o\n";
     $string = "FAILED";
     $container = $o;

     if(strstr($container,$string)) {
       $failed=1;
     } else {
     }
   }

  if($failed){ print "\n\nSorry, something seems to have failed.\n\n"; exit();} 

  $result = mysql_query($query);
  $row = mysql_fetch_array($result);
  $entry_key = $row[0];
  mysql_free_result($result);

print " New entry key: $entry_key\n\n";

#------------------ viper table ---------------------
#$pdb = field_input("PDB ID", NULL);
$pdb = field_input("PDB ID", $vdb);
$name = field_input("Name", fetchName($entry_key));
$pubmed = field_input("Pubmed ID", fetchCitation($entry_key, $pdb));

update_STRUCT_REF($entry_key);
$taxa = fetchTaxa($entry_key);
$taxinfo = split_taxastring($taxa);

if(isset($taxinfo['genome']))
  $genome = field_input("\nGenome", $taxinfo['genome']);
else {
  print " ... no accession code or server error ...";
  $genome = field_input("\nGenome", "NA");
}  

if(isset($taxinfo['family']))
$family = field_input("Family", $taxinfo['family']);
else $family = field_input("Family", "NA");

if(isset($taxinfo['genus']))
$genus = field_input("Genus", $taxinfo['genus']);
else $genus = field_input("Genus", "NA");

print "Enter comma separated list of know hosts for this entry\n";
print "Possible options include: Plant,Archea,Yeast,Bacteria,Insect,Animal,etc.\n";
$hosts = field_input("Hosts", "NA");

$resolution = field_input("Resolution", fetchResolution($entry_key));

$serotype = field_input("Serotype (0 if unique, else 1)", 0);
if($serotype) {
  //$unique_relative_id = field_input("Unique Relative ID", NULL);
  $unique_relative_id = field_input("Unique Relative ID", "NA");
}
else{
  $unique_relative_id = $vdb;
}

//$lattice = field_input("Lattice", NULL);
$lattice = field_input("Lattice", "NA");
$layers = field_input("Layers", 1);

// Deposition date extracted from PDB website
print "\nDo you want me to get the deposition date from the PDB website?\n";
$fetch_depo_date = field_input("(Y|N)", "Y");
if($fetch_depo_date == "Y"){
   $cmd="./get_deposition_date.sh ".$vdb."|grep '>>'|awk '{print $2}'";
   //print $cmd;
   $date=system($cmd);
}
else {
   print "Then please check http://www.rcsb.org/pdb/explore/explore.do?structureId=$vdb\n";
   $date="NA";
}   
$date=field_input("Deposition Date", $date);

system("clear");
print "\nThank you. Now select an option:\n";
print "  [1] Use viperize to generate PDB to VIPER matrix\n";
print "  [2] Read PDB to VIPER matrix from viper.mtx (file must exist)\n";
print "  [3] Use Unit Matrix\n";
$use_viperize=field_input("default ", 1);

$matrix = array();

switch($use_viperize){
 case 1:
        exec("/var/www/html/Admin/viperize /var/www/html/Viperdb/Entries/pdb/$vdb.pdb", $matrix);
        break;
 case 2:
        exec("/bin/cat viper.mtx", $matrix);
        break;
 case 3:
        exec("/bin/cat unit.mtx", $matrix);
        break;
}        

print "$matrix[0]\n$matrix[1]\n$matrix[2]\n";

$line1 = preg_split("/\s+/", trim($matrix[0]));
$line2 = preg_split("/\s+/", trim($matrix[1]));
$line3 = preg_split("/\s+/", trim($matrix[2]));

print "\nPDB to VIPER matrix used is: \n";

print "$line1[0]  $line1[1]  $line1[2]\n";
 print "$line2[0]  $line2[1]  $line2[2]\n";
 print "$line3[0]  $line3[1]  $line3[2]\n";

 print "$line1[3]  $line2[3]  $line3[3]\n";

print "\nYou might want to check if this matrix gives the right VIPERdb orientation.\nUse it to transform the original PDB coordinates and look at the final structure.\nPress enter to continue. ";
fgets(STDIN);


#-------------------- layer and layer_chain --------------
$tnumbers = array();
$layer_names = array();
$layer_ids = array();
$sub_names = array();
$min_ds = array();
$ave_ds = array();
$max_ds = array();

$chain_key = array();

$is_T3=0;
$correct_T3="N";
$i = 0;
while($i < $layers) {
  $this_layer=$i+1;
  system("clear");
  print "\nInformation for layer $this_layer:\n";
  $tnumbers[$i] = field_input("T-Number (i.e. 1,3,pT3,etc)", 1);
  if($tnumbers[$i]==3)$is_T3=1;
  if($layers > 1) {
    if($this_layer == 1) { 
       $defname = "inner"; 
       $defid   = $vdb."_inner"; 
    }
    else { 
       $defname = "outer"; 
       $defid   = $vdb."_outer"; 
    }
    $layer_names[$i] = field_input("Layer Name (outer/inner)", $defname);
    $layer_ids[$i] = field_input("Layer ID (i.e. 1EJ6_INNER)", $defid);
    $sub_names[$i] = field_input("Subunit Name (i.e. VP7)", "NA");
  }
  
  #print possible chain options
  print "\nAssociate current layer with one or more entities listed below: \n";
  $results = mysql_query("select entity_key, type, pdbx_description, pdbx_fragment from ENTITY where entry_key='$entry_key'");
  $num_rows = mysql_num_rows($results); 
  if($num_rows > 0) {
    $ent_tmp="";
    while($row = mysql_fetch_array($results)) {
      print "-- $row[entity_key] : $row[type], $row[pdbx_description]\n";
      $ent_tmp .= $row['entity_key'] . " ";
    }
    $entities_r = field_input("Entities (select #'s from above)", rtrim($ent_tmp));
    $entities = preg_split("/\s+/", trim($entities_r));
    $chain_key[$i] = $entities;

    $i++;
  }  
  else {
    print "WARNING: Something wrong happened, couldn't find an entity_key for entry_key $entry_key layer $i\n";
  }
}

if($is_T3) $correct_T3 = field_input("Correct for T3 (Y|N)", "N");

if($correct_T3 == "Y"){

# VMAT CORRECTION for T=3 Entries
 $m1[0]= 0.30902; $m1[1]= 0.80902; $m1[2]=-0.50000;
 $m2[0]=-0.80902; $m2[1]= 0.50000; $m2[2]= 0.30902;
 $m3[0]= 0.50000; $m3[1]= 0.30902; $m3[2]= 0.80902; 
# 0.000000000    0.000000000    0.000000000

 $ret1[0] = $m1[0]*$line1[0]+$m1[1]*$line2[0]+$m1[2]*$line3[0];
 $ret1[1] = $m1[0]*$line1[1]+$m1[1]*$line2[1]+$m1[2]*$line3[1];
 $ret1[2] = $m1[0]*$line1[2]+$m1[1]*$line2[2]+$m1[2]*$line3[2];

 $ret2[0] = $m2[0]*$line1[0]+$m2[1]*$line2[0]+$m2[2]*$line3[0];
 $ret2[1] = $m2[0]*$line1[1]+$m2[1]*$line2[1]+$m2[2]*$line3[1];
 $ret2[2] = $m2[0]*$line1[2]+$m2[1]*$line2[2]+$m2[2]*$line3[2];

 $ret3[0] = $m3[0]*$line1[0]+$m3[1]*$line2[0]+$m3[2]*$line3[0];
 $ret3[1] = $m3[0]*$line1[1]+$m3[1]*$line2[1]+$m3[2]*$line3[1];
 $ret3[2] = $m3[0]*$line1[2]+$m3[1]*$line2[2]+$m3[2]*$line3[2];

 $ret1[3] = $m1[0]*$line1[3]+$m1[1]*$line2[3]+$m1[2]*$line3[3];
 $ret2[3] = $m2[0]*$line1[3]+$m2[1]*$line2[3]+$m2[2]*$line3[3];
 $ret3[3] = $m3[0]*$line1[3]+$m3[1]*$line2[3]+$m3[2]*$line3[3];

 $line1[0]=$ret1[0];
 $line1[1]=$ret1[1];
 $line1[2]=$ret1[2];
 $line1[3]=$ret1[3];

 $line2[0]=$ret2[0];
 $line2[1]=$ret2[1];
 $line2[2]=$ret2[2];
 $line2[3]=$ret2[3];

 $line3[0]=$ret3[0];
 $line3[1]=$ret3[1];
 $line3[2]=$ret3[2];
 $line3[3]=$ret3[3];

 print "\n\n-- VMAT CORRECTION done:\n\n";

 print "$line1[0]  $line1[1]  $line1[2]\n";
 print "$line2[0]  $line2[1]  $line2[2]\n";
 print "$line3[0]  $line3[1]  $line3[2]\n";

 print "$line1[3]  $line2[3]  $line3[3]\n";

}
else print "\n\n-- No T=3 VMAT CORRECTION made\n";

print "\n\nPress 'Enter' to continue\n";
fgets(STDIN);

system("clear");
print "\nChecking for changes in chain IDs:\n";

# check if chain id's have changed from cif file (this info is already in the database)
  $chain_ID_changed = 0;
  $q = "select distinct auth_asym_id,label_asym_id,pdbx_description from ATOM_SITE,ENTITY where ATOM_SITE.entry_key = ENTITY.entry_key and label_entity_key = entity_key
   and ATOM_SITE.entry_key = $entry_key and label_asym_id is not NULL";
  $result = mysql_query($q);
  $num_rows = mysql_num_rows($result); 

  echo "\n\n$num_rows rows returned:\n\n";

  while($row = mysql_fetch_array($result)){
    $auth=$row['auth_asym_id'];
    $label=$row['label_asym_id'];
    $descr=$row['pdbx_description'];
    if ($auth != $label) {
      $chain_ID_changed = 1;
      print "\nWARNING!!";
      print "\nOriginal chain ID $auth ($descr) found in PDB HAS BEEN CHANGED in CIF File to $label,\n";
      print "\nNOTE: As a standard procedure, WATER asym_id HAS to be set as 'W' and METALS to 'M'. If this was not done automatically, please do it manually now.\n\n";

      print "What do you want me to do?:\n";
      print "  [1] Change it back to $auth\n";
      print "  [2] Provide your own label\n";
      print "  [3] It's fine, I know what I'm doing.\n";
      $asym_id_changed=field_input("default ", 1);

      switch($asym_id_changed){
         case 1:
              $cmd="./rename_chain.pl $entry_key A $auth $auth";
              system($cmd);
              break;
         case 2:
              $new_label=field_input("Enter new Label ", "W");
              $cmd="./rename_chain.pl $entry_key A $auth $new_label";
              system($cmd);
              break;
         case 3:
              print "I hope you're right... left as is.\n\n";
              break;
      }        
    }
    else {
      print "Chain ID $auth: ($descr)  --No change found\n";
    }
  }

print "\nPress 'Enter' to continue\n";
fgets(STDIN);

#----------------------------------------------------------------------------

system("clear");
#insert entry into database
print "\nWe are ready to insert data into viperdb. Press 'Enter', or 'Ctrl+C' to quit.\n";
fgets(STDIN);

#insert viper
$query = 
"INSERT INTO viper (
  entry_key, entry_id, name, genome, genus, family, host,
  serotype, unique_relative_id, layers, matrix_0_0, matrix_0_1,
  matrix_0_2, matrix_1_0, matrix_1_1, matrix_1_2, matrix_2_0,
  matrix_2_1, matrix_2_2, vector_0, vector_1, vector_2,
  times_viewed,private
)
VALUES(
  '$entry_key', '$vdb', '$name', '$genome', '$genus', '$family','$hosts',
  '$serotype', '$unique_relative_id', '$layers', $line1[0],
  $line1[1], $line1[2], $line2[0], $line2[1], $line2[2], $line3[0],
  $line3[1], $line3[2], $line1[3], $line2[3], $line3[3],
  0,0
)";

$results = mysql_query($query);
if(!$results) {
  print "\n\n\nERROR:   Insert failed. " .mysql_error(). "\n\n\n";
  exit();
}

#update possible NULL fields
if($pdb) {
  $query = "UPDATE viper SET pdb_id='$pdb' where entry_key='$entry_key'";
  mysql_query($query);
}
if($pubmed) {
  $query = "UPDATE viper SET pubmed_id='$pubmed' where entry_key='$entry_key'";
  mysql_query($query);
}
if($lattice) {
  $query = "UPDATE viper SET lattice='$lattice' where entry_key='$entry_key'";
  mysql_query($query);
}
if($resolution) {
  $query = "UPDATE viper SET resolution='$resolution' where entry_key='$entry_key'";
  mysql_query($query);
}
mysql_query("UPDATE MMS_ENTRY SET deposition_date='".$date."' WHERE id='".$vdb."'")
    or die(mysql_error());

#flag entry obsolete, so it doesn't show up until ready
$query = "update viper set supercede='1' where entry_key='$entry_key'";
mysql_query($query);

#insert layer
$i = 0;
while($i < $layers) {
  //Make a third, "complete" layer
  if ($layers == 2) {
    $layers ++;
  }
  
  $this_layer=$i+1;
  if ($i == 2) {
    $query = "INSERT INTO layer (entry_key, entry_id) VALUES ('$entry_key','$vdb')";
    array_push($layer_names,"complete");
    array_push($layer_ids,"$vdb"."_COMPLETE");
    array_push($sub_names,"");
  }
  else {
    $query = "INSERT INTO layer (entry_key, entry_id, tnumber) VALUES('$entry_key','$vdb','$tnumbers[$i]')";
  }
  //On the complete layer, insert only a certain amount of information
  mysql_query($query);

  #find layer key
  #The ordering is required to ensure that layers are given unique keys.
  $query = "select layer_key from layer where entry_key='$entry_key' order by layer_key DESC";
  $result = mysql_query($query);
  $row = mysql_fetch_array($result);
  $layer_key = $row[0];

  #possible null fields
  if(isset($layer_names[$i])) {
    $query = "UPDATE layer SET layer_name='$layer_names[$i]' where entry_key='$entry_key' && layer_key=$layer_key";
    mysql_query($query);
    $layer[$i]=$layer_names[$i];
  }
  if(isset($layer_ids[$i])) {
    $query = "UPDATE layer SET layer_id='$layer_ids[$i]' where entry_key='$entry_key' && layer_key=$layer_key";
    mysql_query($query);
   # $layer[$i]=$layer_ids[$i];
  }
  if(isset($sub_names[$i])) {
    $query = "UPDATE layer SET subunit_name='$sub_names[$i]' where entry_key='$entry_key' && layer_key=$layer_key";
    mysql_query($query);
   # $layer[$i]=$sub_names[$i];
  }

  #insert layer_chain
  if($i != 2) {
    $j = 0;
    while($j < count($chain_key[$i])) {
      $ch = $chain_key[$i][$j];
      $query = "INSERT INTO layer_chain (layer_chain_key, layer_key,
        entry_key, entry_id)
        VALUES($ch, $layer_key, $entry_key, '$vdb')";
      //print "$query\n";
      //print_r($chain_key);
      //print_r($chain_key[$i][$j]);
      mysql_query($query);
      $j++;
    }
  }

  print "\nCalculating diameters";

  #create the vdb file for this layer
  #$outfile = "/var/www/html/VDB/$vdb.vdb";
  #$outfile = "/var/www/html/VDB/$vdb"."_"."$layer[$i].vdb";
  $outfile = "/var/www/html/VDB/$vdb.vdb";
  if($layers > 1 && ($i != 2)) {
    echo "\n\nlayer_key=$layer_key was assigned to layer $layer[$i]. \n\n";
    $outfile = "/var/www/html/VDB/$vdb"."_"."$layer[$i].vdb";
    system("/var/www/html/Viperdb/$database"."_prot_na_pdb.pl -entry $vdb -layer $layer[$i] -o $outfile");
  }
  else {
    system("/var/www/html/Viperdb/$database"."_prot_na_pdb.pl -entry $vdb -o $outfile");
  }

  unset($diameters_r);
  exec("/var/www/html/Admin/diameters.pl $outfile", $diameters_r);
  $diameters = explode(" ", trim($diameters_r[0]));
  $min_ds[$i] = $diameters[0];
  $ave_ds[$i] = $diameters[1];
  $max_ds[$i] = $diameters[2];
  #echo "\n\nLayer $layer_ids[$i] in $outfile: Min $min_ds[$i], AVG $ave_ds[$i], Max $max_ds[$i]. Updating db with layer_key=$layer_key.\n\n";
  echo "\n\nLayer in $outfile: Min $min_ds[$i], AVG $ave_ds[$i], Max $max_ds[$i]. Updating db with layer_key=$layer_key.\n\n";
  $query = "UPDATE layer SET min_diameter='$min_ds[$i]', ave_diameter='$ave_ds[$i]', max_diameter='$max_ds[$i]' where entry_key='$entry_key' && layer_key=$layer_key";
  mysql_query($query);
  $i++;
}

# reset the amount of layers. 
  if ($layers == 3){
    echo"layercheck";
   # $layers = 2;
  }

# find out how many and what are the chains  
  $q = "select distinct label_asym_id from ATOM_SITE where ATOM_SITE.entry_key = $entry_key and label_asym_id is not NULL";
  $result = mysql_query($q);
  $num_chains = mysql_num_rows($result); 

  echo "\n\n$num_chains chains found\n\n";

  $list_of_chains="";
  $my_chains = array();
  $i=0;
  while($row = mysql_fetch_array($result)){
    $i++;
    $my_chains[$i]=$row['label_asym_id'];
    $list_of_chains .= $my_chains[$i] . ", ";
    #print "$i, $my_chains[$i]";
  }  
  #print $list_of_chains;

      $chain_moved=0;
      $move_chain = "Y";
      $move_all = "Y";
      while($move_chain=="Y" || $move_chain=="y"){
         $move_chain = field_input("Do you want to move a chain before proceding? (Y|N)", "N");
         if($move_chain=="Y" || $move_chain=="y"){
            $chain_moved=1;

            if($move_all=="Y" || $move_all=="y"){
               $move_all = field_input("Do you want to move all chains at once? (Y|N)", "N");
               if($move_all=="N"  || $move_all=="n") {
                  print "\n Continue with individual chains \n\n";
               }
               else if($move_all=="Y" || $move_all=="y") {
                  $new_place = field_input("Choose Transformation Matrix (1-60) ", "1");
                      print "Moving chain ";
                  for($i=1;$i<=$num_chains;$i++) {
                      print "$my_chains[$i], ";
                      $cmd="./move_chain.pl $vdb $my_chains[$i] $new_place";
                      system($cmd);
                  }    
                  print "\n\nDone moving all chains.\n\n";
                  break;
               }
               else {
                  print "  ERROR!! I dont recognize option $move_all\n\n\n";
                  exit;
               }
            }

            $chain_to_move = field_input("Input Chain ID ($list_of_chains) ", "A");
            $new_place = field_input("Choose Transformation Matrix (1-60) ", "1");
            $cmd="./move_chain.pl $vdb $chain_to_move $new_place";
            system($cmd);
            print "Done moving chain $chain_to_move using transformation matrix $new_place.\n\n";

         }
      }

#  system("clear");
  if($chain_moved){
     print "\nI need to re-create the vdb file\n";
     $outfile = "/var/www/html/VDB/$vdb.vdb";
     system("/var/www/html/Viperdb/$database"."_prot_na_pdb.pl -entry $vdb -o $outfile");
  }   

  $skip_entry_analysis = field_input("Do you want to skip the analysis for: $vdb (Y|N)", "N");
  $cmd="./new_entry.awk $vdb ";

  for ($i = 0; $i < ($layers-1); $i++) {
      $cmd .= $vdb.'_'.$layer_names[$i]." ";
  }
  if (strtolower($skip_entry_analysis) == 'y') {
      $cmd .= "yes";
      print "\n\n Press 'Enter' to proceed with image creation ($cmd) or 'Ctrl+C' to quit. \n\n";
  }
  else {
      print "\n\n Press 'Enter' to procede with analysis ($cmd) or 'Ctrl+C' to quit.\n\n";
  }
  fgets(STDIN);
  system($cmd);
  fgets(STDIN);
  system("clear");

  # -------------
  # blur.sh and do_image are run once for every kind of entry
  # -------------
#  fgets(STDIN);
#  #system("cd "$admin_base_dir);
#  system("clear");
#  $cmd="./blur.sh $vdb";
#  print "\n\n Press 'Enter' to proceed with image creation ($cmd) or 'Ctrl+C' to quit.\n\n";
#  fgets(STDIN);
#  system($cmd);
#
#  fgets(STDIN);
#  #system("cd "$admin_base_dir);
#  system("clear");
#  $cmd="./do_image $vdb";
#  print "\n\n Press 'Enter' to proceed with image creation ($cmd) or 'Ctrl+C' to quit.\n\n";
#  fgets(STDIN);
#  system($cmd);

    $i=0;
    $current_layer = $vdb;

    while ($i < $layers){
        if ($layer_names && ($layer_names[$i] == 'complete')) {
            $current_layer = $vdb;
        }
        else {
            $current_layer = $vdb." ".$layer_names[$i];
        }
        system("clear");
        $cmd = "./blur_v2.sh $current_layer";
        print "\n\n Press 'Enter' to proceed with layer image creation ($cmd) or 'Ctrl+C' to quit.\n\n";
        fgets(STDIN);
        system($cmd);

        fgets(STDIN);
        system("clear");
        $cmd = "./do_image_v2 $current_layer";
        print "\n\n Press 'Enter' to proceed with layer image creation ($cmd) or 'Ctrl+C' to quit.\n\n";
        fgets(STDIN);
        system($cmd);

        $i++;
    }

  $is_user_submitted = field_input("Is this an User Submitted Entry? (Y|N)", "N");
  if($is_user_submitted=="N" || $is_user_submitted=="n"){
     $query = "UPDATE viper SET supercede=NULL where entry_id='$vdb'";
     mysql_query($query);
     print "\nDB 'supercede' has been updated to NULL\n";  
  }
  else print "\nNo need to update DB 'supercede'\n";

  print "\n\n That's all.\n\n";
?>