getmimarkspackagecommand.cpp

   1 //
   2 //  getmimarkspackagecommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 3/25/14.
   6 //  Copyright (c) 2014 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "getmimarkspackagecommand.h"
  10 #include "groupmap.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> GetMIMarksPackageCommand::setParameters(){
  14         try {
  15         //files that have dependancies
  16         CommandParameter pgroup("group", "InputTypes", "", "", "groupOligos", "none", "none","",false,false); parameters.push_back(pgroup);
  17         CommandParameter pfile("file", "InputTypes", "", "", "groupOligos", "none", "none","",false,false); parameters.push_back(pfile);
  18         CommandParameter poligos("oligos", "InputTypes", "", "", "groupOligos", "none", "none","",false,false); parameters.push_back(poligos);
  19         CommandParameter ppackage("package", "Multiple", "air-host_associated-human_associated-human_gut-human_oral-human_skin-human_vaginal-microbial-miscellaneous-plant_associated-sediment-soil-wastewater-water", "miscellaneous", "", "", "","",false,false,true); parameters.push_back(ppackage);
  20         CommandParameter prequiredonly("requiredonly", "Boolean", "", "F", "", "", "","",false,false, true); parameters.push_back(prequiredonly);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "GetMIMarksPackageCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string GetMIMarksPackageCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The get.mimarkspackage command creates a mimarks package form with your groups. The required fields are flagged with * characters. \n";
  38         helpString += "Further documentation on the different packages and required formats can be found here, http://www.mothur.org/wiki/MIMarks_Data_Packages.\n";
  39                 helpString += "The get.mimarkspackage command parameters are: oligos, group, package and requiredonly. oligos or group is required.\n";
  40                 helpString += "The oligos parameter is used to provide your oligos file so mothur can extract your group names.\n";
  41         helpString += "The group parameter is used to provide your group file so mothur can extract your group names.\n";
  42         helpString += "The package parameter is used to select the mimarks package you would like to use. The choices are: air, host_associated, human_associated, human_gut, human_oral, human_skin, human_vaginal, microbial, miscellaneous, plant_associated, sediment, soil, wastewater or waterc. Default=miscellaneous.\n";
  43         helpString += "The requiredonly parameter is used to indicate you only want the required mimarks feilds printed. Default=F.\n";
  44                 helpString += "The get.mimarkspackage command should be in the following format: get.mimarkspackage(oligos=yourOligosFile, package=yourPackage)\n";
  45                 helpString += "get.mimarkspackage(oligos=GQY1XT001.oligos, package=human_gut)\n";
  46                 return helpString;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "GetMIMarksPackageCommand", "getHelpString");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 string GetMIMarksPackageCommand::getOutputPattern(string type) {
  55     try {
  56         string pattern = "";
  57
  58         if (type == "tsv") {  pattern = "[filename],tsv"; }
  59         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  60
  61         return pattern;
  62     }
  63     catch(exception& e) {
  64         m->errorOut(e, "GetMIMarksPackageCommand", "getOutputPattern");
  65         exit(1);
  66     }
  67 }
  68 //**********************************************************************************************************************
  69 GetMIMarksPackageCommand::GetMIMarksPackageCommand(){
  70         try {
  71                 abort = true; calledHelp = true;
  72                 setParameters();
  73         vector<string> tempOutNames;
  74                 outputTypes["tsv"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "GetMIMarksPackageCommand", "GetMIMarksPackageCommand");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82 GetMIMarksPackageCommand::GetMIMarksPackageCommand(string option)  {
  83         try {
  84
  85                 abort = false; calledHelp = false;
  86
  87                 //allow user to run help
  88                 if(option == "help") { help(); abort = true; calledHelp = true; }
  89                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  90
  91                 else {
  92                         //valid paramters for this command
  93                         vector<string> myArray = setParameters();
  94
  95                         OptionParser parser(option);
  96                         map<string,string> parameters = parser.getParameters();
  97
  98                         ValidParameters validParameter;
  99                         map<string,string>::iterator it;
 100                         //check to make sure all parameters are valid for command
 101                         for (it = parameters.begin(); it != parameters.end(); it++) {
 102                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 103                         }
 104
 105             vector<string> tempOutNames;
 106                         outputTypes["tsv"] = tempOutNames;
 107
 108                         //if the user changes the input directory command factory will send this info to us in the output parameter
 109                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 110                         if (inputDir == "not found"){   inputDir = "";          }
 111                         else {
 112
 113                                 string path;
 114                                 it = parameters.find("oligos");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 120                                 }
 121
 122                                 it = parameters.find("group");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 128                                 }
 129
 130                 it = parameters.find("file");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                 }
 136
 137             }
 138
 139                         groupfile = validParameter.validFile(parameters, "group", true);
 140                         if (groupfile == "not open") {  groupfile = "";  abort = true; }
 141                         else if (groupfile == "not found") { groupfile = ""; }
 142             else {  m->setGroupFile(groupfile); inputfile = groupfile; }
 143
 144             file = validParameter.validFile(parameters, "file", true);
 145                         if (file == "not open") {  file = "";  abort = true; }
 146                         else if (file == "not found") { file = ""; }
 147             else {  inputfile = file; }
 148
 149             oligosfile = validParameter.validFile(parameters, "oligos", true);
 150                         if (oligosfile == "not found")      {   oligosfile = "";        }
 151                         else if(oligosfile == "not open")       {       abort = true;           }
 152                         else {  m->setOligosFile(oligosfile); inputfile = oligosfile; }
 153
 154             if ((groupfile != "") && (oligosfile != "") && (file != "")) {
 155                 m->mothurOut("[ERROR]: You may not use a group file, file and an oligos file, only one."); m->mothurOutEndLine(); abort = true;
 156             }
 157
 158             if ((groupfile == "") && (oligosfile == "") && (file == "")) {
 159                 oligosfile = m->getOligosFile();
 160                 if (oligosfile != "") { inputfile = oligosfile;  m->mothurOut("Using " + oligosfile + " as input file for the oligos parameter."); m->mothurOutEndLine(); }
 161                 else {
 162                     groupfile = m->getGroupFile();
 163                     if (groupfile != "") { inputfile = groupfile;  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 164                     else {
 165                         m->mothurOut("[ERROR]: You must provide file, groupfile or oligos file for the get.mimarkspackage command."); m->mothurOutEndLine(); abort = true;
 166                     }
 167                 }
 168             }
 169
 170             package = validParameter.validFile(parameters, "package", false);         if (package == "not found") { package = "miscellaneous"; }
 171
 172             if ((package == "air") || (package == "host_associated") || (package == "human_associated") || (package == "human_gut") || (package == "human_oral") || (package == "human_skin") || (package == "human_vaginal") || (package == "microbial") || (package == "miscellaneous") || (package == "plant_associated") || (package == "sediment") || (package == "soil") || (package == "wastewater") || (package == "water")) {}
 173             else {
 174                 m->mothurOut("[ERROR]: " + package + " is not a valid package selection. Choices are: air, host_associated, human_associated, human_gut, human_oral, human_skin, human_vaginal, microbial, miscellaneous, plant_associated, sediment, soil, wastewater or water. Aborting.\n."); abort = true;
 175             }
 176
 177             string temp;
 178                         temp = validParameter.validFile(parameters, "requiredonly", false);     if(temp == "not found"){        temp = "F";     }
 179                         requiredonly = m->isTrue(temp);
 180                 }
 181
 182         }
 183         catch(exception& e) {
 184                 m->errorOut(e, "GetMIMarksPackageCommand", "GetMIMarksPackageCommand");
 185                 exit(1);
 186         }
 187 }
 188 //**********************************************************************************************************************
 189
 190 int GetMIMarksPackageCommand::execute(){
 191         try {
 192
 193                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 194
 195         if (oligosfile != "") { readOligos();   }
 196         else if (file != "")  { readFile();     }
 197         else {  GroupMap groupmap(groupfile); groupmap.readMap(); Groups = groupmap.getNamesOfGroups(); }
 198
 199         for (set<string>::iterator it = uniqueNames.begin(); it != uniqueNames.end(); it++) {  Groups.push_back(*it); }
 200
 201         if (outputDir == "") { outputDir += m->hasPath(inputfile); }
 202         map<string, string> variables;
 203                 variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
 204                 string outputFileName = getOutputFileName("tsv", variables);
 205
 206         ofstream out;
 207                 m->openOutputFile(outputFileName, out);
 208                 outputNames.push_back(outputFileName); outputTypes["tsv"].push_back(outputFileName);
 209
 210         out << "#This is a tab-delimited file. Additional Documentation can be found at http://www.mothur.org/wiki/MIMarks_Data_Packages." << endl;
 211         out << "#Please fill all the required fields indicated with '*'" << endl;
 212         out << "#Unknown or inapplicable fields can be assigned NA value." << endl;
 213         out << "#You may add extra custom fields to this template. Make sure all the fields are separated by tabs." << endl;
 214         out << "#You may remove any fields not required (marked with '*'). Make sure all the fields are separated by tabs." << endl;
 215         out << "#You can edit this template using Microsoft Excel or any other editor. But while saving the file please make sure to save them as 'TAB-DELIMITED' TEXT FILE." << endl;
 216
 217         if (package == "air") {
 218             out << "#Environmental:MIMARKS.specimen.air.3.0" << endl;
 219             if (requiredonly) {
 220                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *altitude" << endl;
 221             }else {
 222                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *altitude       barometric_press        carb_dioxide    carb_monoxide   chem_administration     elev    humidity        methane misc_param      organism_count  oxygen  oxy_stat_samp   perturbation    pollutants      resp_part_matter        samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp solar_irradiance        temp    ventilation_rate        ventilation_type        volatile_org_comp       wind_direction  wind_speed" << endl;
 223             }
 224         }else if (package == "host_associated") {
 225             out << "#Environmental:MIMARKS.specimen.host-associated.3.0" << endl;
 226             if (requiredonly) {
 227                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host   " << endl;
 228             }else {
 229                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods    rel_to_oxygen        samp_collect_device     samp_mat_process        *host   age     altitude        blood_press_diast       blood_press_syst        body_habitat    body_product    tissue  chem_administration     depth   diet    disease_stat    dry_mass        elev    family_relationship     genotype        gravidity       height_or_length        host_body_temp  host_color      host_growth_cond        host_shape      host_subject_id host_taxid      infra_specific_name     infra_specific_rank     last_meal       life_stage      misc_param      organism_count  oxy_stat_samp   perturbation    phenotype       samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     substrate       temp    tot_mass" << endl;
 230             }
 231         }else if (package == "human_associated") {
 232             out << "#Environmental:MIMARKS.specimen.human-associated.3.0" << endl;
 233             if (requiredonly) {
 234                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 235             }else {
 236                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   hiv_stat        ihmc_ethnicity  ihmc_medication_code    age     amniotic_fluid_color    foetal_health_stat      gestation_state maternal_health_stat    blood_blood_disord      body_product    tissue  body_mass_index chem_administration     diet    disease_stat    drug_usage      family_relationship     genotype        height  host_body_temp  host_subject_id last_meal       nose_throat_disord      pulmonary_disord        diet_last_six_month     medic_hist_perform      misc_param      occupation      organism_count  oxy_stat_samp   perturbation    phenotype       pet_farm_animal pulse   samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     smoker  study_complt_stat       temp    tot_mass        travel_out_six_month    twin_sibling    urine_collect_meth      kidney_disord   urogenit_tract_disor    weight_loss_3_month" << endl;
 237             }
 238         }else if (package == "human_gut") {
 239             out << "#Environmental:MIMARKS.specimen.human-gut.3.0" << endl;
 240             if (requiredonly) {
 241                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 242             }else {
 243                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   ihmc_ethnicity  ihmc_medication_code    age     body_product    tissue  body_mass_index chem_administration     diet    disease_stat    family_relationship     gastrointest_disord     genotype        height  host_body_temp  host_subject_id last_meal       liver_disord    medic_hist_perform      misc_param      occupation      organism_count  oxy_stat_samp   perturbation    phenotype       pulse   samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     special_diet    temp    tot_mass" << endl;
 244             }
 245         }else if (package == "human_oral") {
 246             out << "#Environmental:MIMARKS.specimen.human-oral.3.0" << endl;
 247             if (requiredonly) {
 248                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 249             }else {
 250                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   ihmc_ethnicity  ihmc_medication_code    age     body_product    tissue  body_mass_index chem_administration     diet    disease_stat    family_relationship     genotype        height  host_body_temp  host_subject_id last_meal       medic_hist_perform      misc_param      nose_mouth_teeth_throat_disord  occupation      organism_count  oxy_stat_samp   perturbation    phenotype       pulse   samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     temp    time_last_toothbrush    tot_mass" << endl;
 251             }
 252         }else if (package == "human_skin") {
 253             out << "#Environmental:MIMARKS.specimen.human-skin.3.0" << endl;
 254             if (requiredonly) {
 255                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 256             }else {
 257                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   ihmc_ethnicity  ihmc_medication_code    age     body_product    tissue  body_mass_index chem_administration     dermatology_disord      diet    disease_stat    dominant_hand   family_relationship     genotype        height  host_body_temp  host_subject_id last_meal       medic_hist_perform      misc_param      occupation      organism_count  oxy_stat_samp   perturbation    phenotype       pulse   samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     temp    time_since_last_wash    tot_mass" << endl;
 258             }
 259         }else if (package == "human_vaginal") {
 260             out << "#Environmental:MIMARKS.specimen.human-vaginal.3.0" << endl;
 261             if (requiredonly) {
 262                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 263             }else {
 264                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   hrt     ihmc_ethnicity  ihmc_medication_code    age     birth_control   body_product    tissue  body_mass_index chem_administration     diet    disease_stat    douche  family_relationship     genotype        gynecologic_disord      height  host_body_temp  host_subject_id hysterectomy    last_meal       medic_hist_perform      menarche        menopause       misc_param      occupation      organism_count  oxy_stat_samp   perturbation    phenotype       pregnancy       pulse   samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp sex     sexual_act      temp    tot_mass        urogenit_disord" << endl;
 265             }
 266         }else if (package == "microbial") {
 267             out << "#Environmental:MIMARKS.specimen.microbial.3.0" << endl;
 268             if (requiredonly) {
 269                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *depth  *elev" << endl;
 270             }else {
 271                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *depth  *elev   alkalinity      alkyl_diethers  altitude        aminopept_act   ammonium        bacteria_carb_prod      biomass bishomohopanol  bromide calcium carb_nitro_ratio        chem_administration     chloride        chlorophyll     diether_lipids  diss_carb_dioxide       diss_hydrogen   diss_inorg_carb diss_org_carb   diss_org_nitro  diss_oxygen     glucosidase_act magnesium       mean_frict_vel  mean_peak_frict_vel     methane misc_param      n_alkanes       nitrate nitrite nitro   org_carb        org_matter      org_nitro       organism_count  oxy_stat_samp   ph      part_org_carb   perturbation    petroleum_hydrocarb     phaeopigments   phosphate       phosplipid_fatt_acid    potassium       pressure        redox_potential salinity        samp_size       samp_store_dur  samp_store_loc  samp_store_temp silicate        sodium  sulfate sulfide temp    tot_carb        tot_nitro       tot_org_carb    turbidity       water_content" << endl;
 272             }
 273         }else if (package == "miscellaneous") {
 274             out << "#Environmental:MIMARKS.specimen.miscellaneous.3.0" << endl;
 275             if (requiredonly) {
 276                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *title  *seq_methods    *lat_lon" << endl;
 277             }else {
 278                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        alkalinity      altitude        ammonium        biomass bromide calcium chem_administration     chloride        chlorophyll     current density depth   diether_lipids  diss_carb_dioxide       diss_hydrogen   diss_inorg_carb diss_org_nitro  diss_oxygen     elev    misc_param      nitrate nitrite nitro   org_carb        org_matter      org_nitro       organism_count  oxy_stat_samp   ph      perturbation    phosphate       phosplipid_fatt_acid    potassium       pressure        salinity        samp_size       samp_store_dur  samp_store_loc  samp_store_temp silicate        sodium  sulfate sulfide temp" << endl;
 279             }
 280         }else if (package == "plant_associated") {
 281             out << "#Environmental:MIMARKS.specimen.plant-associated.3.0" << endl;
 282             if (requiredonly) {
 283                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *host" << endl;
 284             }else {
 285                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *host   age     air_temp_regm   altitude        antibiotic_regm body_product    chem_administration     chem_mutagen    climate_environment     depth   disease_stat    dry_mass        elev    fertilizer_regm fungicide_regm  gaseous_environment     genotype        gravity growth_hormone_regm     growth_med      height_or_length        herbicide_regm  host_taxid      humidity_regm   infra_specific_name     infra_specific_rank     life_stage      mechanical_damage       mineral_nutr_regm       misc_param      non_mineral_nutr_regm   organism_count  oxy_stat_samp   ph_regm perturbation    pesticide_regm  phenotype       tissue  plant_product   radiation_regm  rainfall_regm   salt_regm       samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp season_environment      standing_water_regm     temp    tiss_cult_growth_med    tot_mass        water_temp_regm watering_regm   wet_mass" << endl;
 286             }
 287         }else if (package == "sediment") {
 288             out << "#Environmental:MIMARKS.specimen.sediment.3.0" << endl;
 289             if (requiredonly) {
 290                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *depth  *elev" << endl;
 291             }else {
 292                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *depth  *elev   alkalinity      alkyl_diethers  aminopept_act   ammonium        bacteria_carb_prod      biomass bishomohopanol  bromide calcium carb_nitro_ratio        chem_administration     chloride        chlorophyll     density diether_lipids  diss_carb_dioxide       diss_hydrogen   diss_inorg_carb diss_org_carb   diss_org_nitro  diss_oxygen     glucosidase_act magnesium       mean_frict_vel  mean_peak_frict_vel     methane misc_param      n_alkanes       nitrate nitrite nitro   org_carb        org_matter      org_nitro       organism_count  oxy_stat_samp   ph      particle_class  part_org_carb   perturbation    petroleum_hydrocarb     phaeopigments   phosphate       phosplipid_fatt_acid    porosity        potassium       pressure        redox_potential salinity        samp_size       samp_store_dur  samp_store_loc  samp_store_temp sediment_type   silicate        sodium  sulfate sulfide temp    tidal_stage     tot_carb        tot_nitro       tot_org_carb    turbidity       water_content" << endl;
 293             }
 294         }else if (package == "soil") {
 295             out << "#Environmental:MIMARKS.specimen.soil.3.0" << endl;
 296             if (requiredonly) {
 297                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *depth  *elev" << endl;
 298             }else {
 299                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *depth  *elev   altitude        sieving cur_land_use    cur_vegetation_meth     cur_vegetation  drainage_class  al_sat  al_sat_meth     heavy_metals_meth       heavy_metals    salinity_meth   extreme_salinity        fao_class       agrochem_addition       crop_rotation   extreme_event   fire    flooding        previous_land_use_meth  previous_land_use       tillage horizon_meth    horizon link_class_info link_climate_info       link_addit_analys       annual_season_precpt    annual_season_temp      microbial_biomass_meth  microbial_biomass       misc_param      other   ph_meth ph      pool_dna_extracts       profile_position        samp_size       samp_weight_dna_ext     slope_aspect    slope_gradient  soil_type_meth  soil_type       local_class_meth        local_class     store_cond      texture_meth    texture tot_n_meth      tot_n   tot_org_c_meth  tot_org_carb    water_content_soil_meth water_content_soil" << endl;
 300             }
 301         }else if (package == "wastewater") {
 302             out << "#Environmental:MIMARKS.specimen.wastewater.3.0" << endl;
 303             if (requiredonly) {
 304                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods" << endl;
 305             }else {
 306                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        alkalinity      biochem_oxygen_dem      chem_administration     chem_oxygen_dem depth   efficiency_percent      emulsions       gaseous_substances      indust_eff_percent      inorg_particles misc_param      nitrate org_particles   organism_count  oxy_stat_samp   ph      perturbation    phosphate       pre_treatment   primary_treatment       reactor_type    samp_size       samp_salinity   samp_store_dur  samp_store_loc  samp_store_temp secondary_treatment     sewage_type     sludge_retent_time      sodium  soluble_inorg_mat       soluble_org_mat suspend_solids  temp    tertiary_treatment      tot_nitro       tot_phosphate   wastewater_type" << endl;
 307             }
 308         }else if (package == "water") {
 309             out << "#Environmental:MIMARKS.specimen.water.3.0" << endl;
 310             if (requiredonly) {
 311                 out << "*sample_name    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods *depth" << endl;
 312             }else {
 313                 out << "*sample_name    description     bioproject_id   sample_title    *organism       *collection_date        *biome  *feature        *material       *geo_loc_name   *lat_lon   *title  *seq_methods rel_to_oxygen   samp_collect_device     samp_mat_process        *depth  alkalinity      alkyl_diethers  aminopept_act   ammonium        atmospheric_data        bacteria_carb_prod      biomass bishomohopanol  bromide calcium carb_nitro_ratio        chem_administration     chloride        chlorophyll     current density diether_lipids  diss_carb_dioxide       diss_hydrogen   diss_inorg_carb diss_inorg_nitro        diss_inorg_phosp        diss_org_carb   diss_org_nitro  diss_oxygen     elev    glucosidase_act light_intensity magnesium       mean_frict_vel  mean_peak_frict_vel     misc_param      n_alkanes       nitrate nitrite nitro   org_carb        org_matter      org_nitro       organism_count  oxy_stat_samp   ph      part_org_carb   part_org_nitro  perturbation    petroleum_hydrocarb     phaeopigments   phosphate       phosplipid_fatt_acid    photon_flux     potassium       pressure        primary_prod    redox_potential salinity        samp_size       samp_store_dur  samp_store_loc  samp_store_temp silicate        sodium  soluble_react_phosp     sulfate sulfide suspend_part_matter     temp    tidal_stage     tot_depth_water_col     tot_diss_nitro  tot_inorg_nitro tot_nitro       tot_part_carb   tot_phosp" << endl;
 314             }
 315         }
 316
 317         for (int i = 0; i < Groups.size(); i++) {  out << Groups[i] << '\t' << endl; }
 318
 319         out.close();
 320
 321         //output files created by command
 322                 m->mothurOutEndLine();
 323                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 324                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 325                 m->mothurOutEndLine();
 326         return 0;
 327
 328     }
 329         catch(exception& e) {
 330                 m->errorOut(e, "GetMIMarksPackageCommand", "execute");
 331                 exit(1);
 332         }
 333 }
 334 //***************************************************************************************************************
 335 int GetMIMarksPackageCommand::readOligos(){
 336         try {
 337                 ifstream inOligos;
 338                 m->openInputFile(oligosfile, inOligos);
 339
 340                 string type, oligo, roligo, group;
 341         vector<string> primerNameVector, barcodeNameVector;
 342         set<string> uniquePrimers;
 343         set<string> uniqueBarcodes;
 344
 345                 while(!inOligos.eof()){
 346
 347                         inOligos >> type;
 348
 349                         if (m->debug) { m->mothurOut("[DEBUG]: reading type - " + type + ".\n"); }
 350
 351                         if(type[0] == '#'){
 352                                 while (!inOligos.eof()) {       char c = inOligos.get();  if (c == 10 || c == 13){      break;  }       } // get rest of line if there's any crap there
 353                                 m->gobble(inOligos);
 354                         }
 355                         else{
 356                                 m->gobble(inOligos);
 357                                 //make type case insensitive
 358                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
 359
 360                                 inOligos >> oligo;
 361
 362                 if (m->debug) { m->mothurOut("[DEBUG]: reading - " + oligo + ".\n"); }
 363
 364                                 for(int i=0;i<oligo.length();i++){
 365                                         oligo[i] = toupper(oligo[i]);
 366                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
 367                                 }
 368
 369                                 if(type == "FORWARD"){
 370                                         group = "";
 371
 372                                         // get rest of line in case there is a primer name
 373                                         while (!inOligos.eof()) {
 374                                                 char c = inOligos.get();
 375                                                 if (c == 10 || c == 13 || c == -1){     break;  }
 376                                                 else if (c == 32 || c == 9){;} //space or tab
 377                                                 else {  group += c;  }
 378                                         }
 379
 380                                         primerNameVector.push_back(group);
 381                                 }
 382                 else if (type == "PRIMER"){
 383                     m->gobble(inOligos);
 384
 385                     inOligos >> roligo;
 386
 387                     for(int i=0;i<roligo.length();i++){
 388                         roligo[i] = toupper(roligo[i]);
 389                         if(roligo[i] == 'U')    {       roligo[i] = 'T';        }
 390                     }
 391
 392                     group = "";
 393
 394                                         // get rest of line in case there is a primer name
 395                                         while (!inOligos.eof()) {
 396                                                 char c = inOligos.get();
 397                                                 if (c == 10 || c == 13 || c == -1){     break;  }
 398                                                 else if (c == 32 || c == 9){;} //space or tab
 399                                                 else {  group += c;  }
 400                                         }
 401
 402                                         primerNameVector.push_back(group);
 403                 }else if(type == "BARCODE"){
 404                                         inOligos >> group;
 405
 406                     //barcode lines can look like   BARCODE   atgcatgc   groupName  - for 454 seqs
 407                     //or                            BARCODE   atgcatgc   atgcatgc    groupName  - for illumina data that has forward and reverse info
 408
 409                     string temp = "";
 410                     while (!inOligos.eof())     {
 411                                                 char c = inOligos.get();
 412                                                 if (c == 10 || c == 13 || c == -1){     break;  }
 413                                                 else if (c == 32 || c == 9){;} //space or tab
 414                                                 else {  temp += c;  }
 415                                         }
 416
 417                     //then this is illumina data with 4 columns
 418                     if (temp != "") {
 419
 420                         string reverseBarcode = group; //reverseOligo(group); //reverse barcode
 421                         group = temp;
 422
 423                         barcodeNameVector.push_back(group);
 424                     }else {
 425                         barcodeNameVector.push_back(group);
 426                     }
 427                                 }
 428                         }
 429                         m->gobble(inOligos);
 430                 }
 431                 inOligos.close();
 432
 433                 //add in potential combos
 434                 if(barcodeNameVector.size() == 0){
 435                         barcodeNameVector.push_back("");
 436                 }
 437
 438                 if(primerNameVector.size() == 0){
 439                         primerNameVector.push_back("");
 440                 }
 441
 442
 443         for(int i = 0; i <  barcodeNameVector.size(); i++){
 444             for(int j = 0; j < primerNameVector.size(); j++){
 445
 446                 string primerName = primerNameVector[j];
 447                 string barcodeName = barcodeNameVector[i];
 448
 449                 if ((primerName == "ignore") || (barcodeName == "ignore")) { } //do nothing
 450                 else if ((primerName == "") && (barcodeName == "")) { }
 451                 else {
 452                     string comboGroupName = "";
 453
 454                     if(primerName == ""){
 455                         comboGroupName = barcodeNameVector[i];
 456                     }
 457                     else{
 458                         if(barcodeName == ""){
 459                             comboGroupName = primerNameVector[j];
 460                         }
 461                         else{
 462                             comboGroupName = barcodeNameVector[i] + "." + primerNameVector[j];
 463                         }
 464                     }
 465                     uniqueNames.insert(comboGroupName);
 466                 }
 467             }
 468         }
 469
 470
 471
 472         if (m->debug) { int count = 0; for (set<string>::iterator it = uniqueNames.begin(); it != uniqueNames.end(); it++) { m->mothurOut("[DEBUG]: " + toString(count) + " groupName = " + *it + "\n"); count++; } }
 473
 474
 475                 return true;
 476
 477         }
 478         catch(exception& e) {
 479                 m->errorOut(e, "GetMIMarksPackageCommand", "readOligos");
 480                 exit(1);
 481         }
 482 }
 483 //**********************************************************************************************************************
 484 // going to have to rework this to allow for other options --
 485 /*
 486  file option 1
 487
 488  sfffile1   oligosfile1
 489  sfffile2   oligosfile2
 490  ...
 491
 492  file option 2
 493
 494  fastqfile1 oligosfile1
 495  fastqfile2 oligosfile2
 496  ...
 497
 498  file option 3
 499
 500  fastqfile  fastqfile   group
 501  fastqfile  fastqfile   group
 502  fastqfile  fastqfile   group
 503  ...
 504
 505  */
 506
 507 int GetMIMarksPackageCommand::readFile(){
 508         try {
 509         //vector<string> theseFiles;
 510         inputfile = file;
 511
 512         ifstream in;
 513         m->openInputFile(file, in);
 514
 515         while(!in.eof()) {
 516
 517             if (m->control_pressed) { return 0; }
 518
 519             string line = m->getline(in);  m->gobble(in);
 520             vector<string> pieces = m->splitWhiteSpace(line);
 521
 522             string group = "";
 523             string thisFileName1, thisFileName2; thisFileName1 = ""; thisFileName2 = "";
 524             if (pieces.size() == 2) {
 525                 thisFileName1 = pieces[0];
 526                 thisFileName2 = pieces[1];
 527             }else if (pieces.size() == 3) {
 528                 thisFileName1 = pieces[1];
 529                 thisFileName2 = pieces[2];
 530                 string group = pieces[0];
 531             }else {
 532                 m->mothurOut("[ERROR]: file lines can be 2 or 3 columns. The 2 column files are sff file then oligos or fastqfile then oligos. You may have multiple lines in the file.  The 3 column files are for paired read libraries. The format is groupName, forwardFastqFile reverseFastqFile. \n"); m->control_pressed = true;
 533             }
 534
 535             if (m->debug) { m->mothurOut("[DEBUG]: group = " + group + ", thisFileName1 = " + thisFileName1 + ", thisFileName2 = " + thisFileName2  + ".\n"); }
 536
 537             //check to make sure both are able to be opened
 538             ifstream in2;
 539             int openForward = m->openInputFile(thisFileName1, in2, "noerror");
 540
 541             //if you can't open it, try default location
 542             if (openForward == 1) {
 543                 if (m->getDefaultPath() != "") { //default path is set
 544                     string tryPath = m->getDefaultPath() + m->getSimpleName(thisFileName1);
 545                     m->mothurOut("Unable to open " + thisFileName1 + ". Trying default " + tryPath); m->mothurOutEndLine();
 546                     ifstream in3;
 547                     openForward = m->openInputFile(tryPath, in3, "noerror");
 548                     in3.close();
 549                     thisFileName1 = tryPath;
 550                 }
 551             }
 552
 553             //if you can't open it, try output location
 554             if (openForward == 1) {
 555                 if (m->getOutputDir() != "") { //default path is set
 556                     string tryPath = m->getOutputDir() + m->getSimpleName(thisFileName1);
 557                     m->mothurOut("Unable to open " + thisFileName1 + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 558                     ifstream in4;
 559                     openForward = m->openInputFile(tryPath, in4, "noerror");
 560                     thisFileName1 = tryPath;
 561                     in4.close();
 562                 }
 563             }
 564
 565             if (openForward == 1) { //can't find it
 566                 m->mothurOut("[WARNING]: can't find " + thisFileName1 + ", ignoring.\n");
 567             }else{  in2.close();  }
 568
 569             int openReverse = 1;
 570
 571             ifstream in3;
 572             openReverse = m->openInputFile(thisFileName2, in3, "noerror");
 573
 574             //if you can't open it, try default location
 575             if (openReverse == 1) {
 576                 if (m->getDefaultPath() != "") { //default path is set
 577                     string tryPath = m->getDefaultPath() + m->getSimpleName(thisFileName2);
 578                     m->mothurOut("Unable to open " + thisFileName2 + ". Trying default " + tryPath); m->mothurOutEndLine();
 579                     ifstream in3;
 580                     openReverse = m->openInputFile(tryPath, in3, "noerror");
 581                     in3.close();
 582                     thisFileName2 = tryPath;
 583                 }
 584             }
 585
 586             //if you can't open it, try output location
 587             if (openReverse == 1) {
 588                 if (m->getOutputDir() != "") { //default path is set
 589                     string tryPath = m->getOutputDir() + m->getSimpleName(thisFileName2);
 590                     m->mothurOut("Unable to open " + thisFileName2 + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 591                     ifstream in4;
 592                     openReverse = m->openInputFile(tryPath, in4, "noerror");
 593                     thisFileName2 = tryPath;
 594                     in4.close();
 595                 }
 596             }
 597
 598             if (openReverse == 1) { //can't find it
 599                 m->mothurOut("[WARNING]: can't find " + thisFileName2 + ", ignoring pair.\n");
 600             }else{  in3.close();  }
 601
 602
 603             if ((pieces.size() == 2) && (openForward != 1) && (openReverse != 1)) { //good pair and sff or fastq and oligos
 604                     oligosfile = thisFileName2;
 605                     if (m->debug) { m->mothurOut("[DEBUG]: about to read oligos\n"); }
 606                     readOligos();
 607             }else if((pieces.size() == 3) && (openForward != 1) && (openReverse != 1)) { //good pair and paired read
 608                 Groups.push_back(group);
 609             }
 610         }
 611         in.close();
 612
 613         inputfile = file;
 614
 615         return 0;
 616     }
 617         catch(exception& e) {
 618                 m->errorOut(e, "GetMIMarksPackageCommand", "readFile");
 619                 exit(1);
 620         }
 621 }
 622 //**********************************************************************************************************************
 623
 624