Taking {UPPER_CASE} out of Mpox pattern field

And fix to "host vaccination status menu" range to     "HostVaccinationStatusMenu"

Author: ddooley

lib/DataHarmonizer.js

@@ -2468,45 +2468,49 @@ class DataHarmonizer {
       let matrixRow = matrix[row];
       matrixRow[col] = change[3]; // prime changed value
 
-      const prevName = col > 0 ? fields[col - 1].title : null;
-      const nextName = fields.length > col + 1 ? fields[col + 1].title : null;
+      const prevName = col > 0 ? fields[col - 1].name : null;
+      const nextName = fields.length > col + 1 ? fields[col + 1].name : null;
 
-      // Match <field>[field unit]
-      if (nextName === field.title + ' unit') {
+      // On [field] where next field is [field unit]
+      if (nextName === field.name + '_unit') {
         if (field.datatype === 'xsd:date') {
           // Transform ISO 8601 date to bin year / month granularity.
+          // This literally resets day or month to /01/ value based on 
+          // granularity, for information accuracy or obfuscation purposes.
           // "day" granularity is taken care of by regular date validation.
           // Don't attempt to reformat x/y/z dates here.
-          const dateGranularity = this.hot.getDataAtCell(row, col + 1);
+          const dateGranularity = this.hot.getDataAtCell(row, col + 1).split(' ')[0];
           // previously had to block x/y/z with change[3].indexOf('/') === -1 &&
           if (dateGranularity === 'year' || dateGranularity === 'month') {
             change[3] = this.setDateChange(dateGranularity, change[3]);
           }
           return;
         }
 
-        // Match <field>[field unit][field bin]
+        /* Deprecating this. Users need to change unit to trigger bin
+        // Match <field> [field]_unit [field]_bin
         const nextNextName =
-          fields.length > col + 2 ? fields[col + 2].title : null;
-        if (nextNextName === field.title + ' bin') {
+          fields.length > col + 2 ? fields[col + 2].name : null;
+        if (nextNextName === field.name + '_bin') {
           matrixRow[col + 1] = this.hot.getDataAtCell(row, col + 1); //prime unit
           matrixRow[col + 2] = this.hot.getDataAtCell(row, col + 2); //prime bin
           this.binChangeTest(matrix, col, fields, 2, triggered_changes);
           return;
         }
+        */
       }
 
-      // Match <field>[field bin]
-      if (nextName === field.title + ' bin') {
+      // On [field] here, where next column is "[field]_bin". 
+      if (nextName === field.name + '_bin') {
         matrixRow[col + 1] = this.hot.getDataAtCell(row, col + 1); //prime bin
         this.binChangeTest(matrix, col, fields, 1, triggered_changes);
         return;
       }
 
-      // Match [field]<field unit>
-      if (field.title === prevName + ' unit') {
-        // Match [field]<field unit>[field bin]
-        if (prevName + ' bin' === nextName) {
+      // On "[field] unit" here, where next column is [field]_bin.
+      if (field.name === prevName + '_unit') {
+        if (prevName + '_bin' === nextName) {
+          console.log("on", prevName + '_unit')
           // trigger reevaluation of bin from field
           matrixRow[col - 1] = this.hot.getDataAtCell(row, col - 1);
           matrixRow[col] = this.hot.getDataAtCell(row, col);
@@ -2563,20 +2567,20 @@ class DataHarmonizer {
 
       // Rules that require a column or two following current one.
       if (fields.length > col + 1) {
-        const nexttitle = fields[col + 1].title;
+        const nexttitle = fields[col + 1].name;
 
         // Rule: for any <x>[x bin] pattern of field names,
         // find and set appropriate bin selection.
-        if (nexttitle === field.title + ' bin') {
+        if (nexttitle === field.name + '_bin') {
           this.binChangeTest(matrix, col, fields, 1, triggered_changes);
         }
         // Rule: for any [x], [x unit], [x bin] series of fields
-        else if (nexttitle === field.title + ' unit') {
+        else if (nexttitle === field.name + '_unit') {
           if (fields[col].datatype === 'xsd:date') {
             //Validate
             for (let row = 0; row < matrix.length; row++) {
               if (!matrix[row][col]) continue;
-              const dateGranularity = matrix[row][col + 1];
+              const dateGranularity = matrix[row][col + 1].split(' ')[0];
               if (dateGranularity === 'year' || dateGranularity === 'month') {
                 matrix[row][col] = this.setDateChange(
                   dateGranularity,
@@ -2666,15 +2670,12 @@ class DataHarmonizer {
             // Host age unit is interpreted by default to be year.
             // If user selects month, value is converted into years for binning.
             // Future solution won't hardcode month / year assumption
+            // ENGLISH ONLY: value might be suffixed with ontology id.
             if (unit) {
-              if (unit === 'month') {
+              if (unit.split(' ',1)[0] === 'month') {
                 number = number / 12;
               }
             }
-            // Force unit to be year if empty.
-            //else {
-            //  triggered_changes.push([rowOffset + parseInt(row), col+1, undefined, 'year']);
-            //}
           }
           // .flatVocabulary is an array of string bin ranges e.g. "10 - 19"
           // if (typeof fields[hotRowBinCol].flatVocabulary !== 'undefined') {
@@ -2699,7 +2700,6 @@ class DataHarmonizer {
           const bin_values = fields[hotRowBinCol].flatVocabulary;
           if (value in bin_values && (!bin_value || bin_value === '')) {
             selection = value;
-            console.log('no bin value', value);
           }
           // If a unit field exists, then set that to metadata too.
           if (binOffset == 2) {

web/templates/canada_covid19/schema.json

@@ -7382,7 +7382,7 @@
       ],
       "any_of": [
         {
-          "range": "host vaccination status menu"
+          "range": "HostVaccinationStatusMenu"
         },
         {
           "range": "NullValueMenu"
@@ -12338,7 +12338,7 @@
           "slot_group": "Host vaccination information",
           "any_of": [
             {
-              "range": "host vaccination status menu"
+              "range": "HostVaccinationStatusMenu"
             },
             {
               "range": "NullValueMenu"

web/templates/canada_covid19/schema.yaml

@@ -1687,7 +1687,7 @@ slots:
     comments: Select the vaccination status of the host from the pick list.
     slot_uri: GENEPIO:0001404
     any_of:
-    - range: host vaccination status menu
+    - range: HostVaccinationStatusMenu
     - range: NullValueMenu
     examples:
     - value: Fully Vaccinated

web/templates/canada_covid19/schema_slots.tsv

@@ -64,7 +64,7 @@ CanCOGeNCovid19			GENEPIO:0001122	Database Identifiers
 	Host Information	Informations sur l'hôte	GENEPIO:0001401	pre-existing conditions and risk factors	preexisting_conditions_and_risk_factors	Conditions préexistantes et facteurs de risque	PreExistingConditionsAndRiskFactorsMenu	NullValueMenu		TRUE							Patient pre-existing conditions and risk factors. <li>Pre-existing condition: A medical condition that existed prior to the current infection. <li>Risk Factor: A variable associated with an increased risk of disease or infection.	Conditions préexistantes et facteurs de risque du patient <li>Condition préexistante : Condition médicale qui existait avant l’infection actuelle <li>Facteur de risque : Variable associée à un risque accru de maladie ou d’infection 	Select all of the pre-existing conditions and risk factors experienced by the host from the pick list. If the desired term is missing, contact the curation team.	Sélectionnez toutes les conditions préexistantes et tous les facteurs de risque de l’hôte à partir de la liste de sélection. S’il manque le terme souhaité, veuillez communiquer avec l’équipe de conservation des données.	Asthma; Pregnancy; Smoking	Asthme; grossesse; tabagisme			pre-existing conditions and risk factors		
 	Host Information	Informations sur l'hôte	GENEPIO:0001402	complications	complications	Complications	ComplicationsMenu	NullValueMenu		TRUE							Patient medical complications that are believed to have occurred as a result of host disease.	Complications médicales du patient qui seraient dues à une maladie de l’hôte	Select all of the complications experienced by the host from the pick list. If the desired term is missing, contact the curation team.	Sélectionnez toutes les complications éprouvées par l’hôte à partir de la liste de sélection. S’il manque le terme souhaité, veuillez communiquer avec l’équipe de conservation des données.	Acute Respiratory Failure; Coma; Septicemia	Insuffisance respiratoire aiguë; coma; septicémie			complications		
 			GENEPIO:0001403	Host vaccination information																							
-	Host vaccination information	Informations sur la vaccination de l'hôte	GENEPIO:0001404	host vaccination status	host_vaccination_status	Statut de vaccination de l’hôte	host vaccination status menu	NullValueMenu									The vaccination status of the host (fully vaccinated, partially vaccinated, or not vaccinated).	Statut de vaccination de l’hôte (entièrement vacciné, partiellement vacciné ou non vacciné)	Select the vaccination status of the host from the pick list.	Sélectionnez le statut de vaccination de l’hôte à partir de la liste de sélection.	Fully Vaccinated	Entièrement vacciné			PH_VACCINATION_HISTORY		
+	Host vaccination information	Informations sur la vaccination de l'hôte	GENEPIO:0001404	host vaccination status	host_vaccination_status	Statut de vaccination de l’hôte	HostVaccinationStatusMenu	NullValueMenu									The vaccination status of the host (fully vaccinated, partially vaccinated, or not vaccinated).	Statut de vaccination de l’hôte (entièrement vacciné, partiellement vacciné ou non vacciné)	Select the vaccination status of the host from the pick list.	Sélectionnez le statut de vaccination de l’hôte à partir de la liste de sélection.	Fully Vaccinated	Entièrement vacciné			PH_VACCINATION_HISTORY		
 	Host vaccination information	Informations sur la vaccination de l'hôte	GENEPIO:0001406	number of vaccine doses received	number_of_vaccine_doses_received	Nombre de doses du vaccin reçues	integer										The number of doses of the vaccine recived by the host.	Nombre de doses du vaccin reçues par l’hôte	Record how many doses of the vaccine the host has received.	Enregistrez le nombre de doses du vaccin que l’hôte a reçues.	2	2					
 	Host vaccination information	Informations sur la vaccination de l'hôte	GENEPIO:0100313	vaccination dose 1 vaccine name	vaccination_dose_1_vaccine_name	Dose du vaccin 1 – Nom du vaccin	VaccineNameMenu	NullValueMenu									The name of the vaccine administered as the first dose of a vaccine regimen.	Nom du vaccin administré comme première dose d’un schéma vaccinal	Provide the name and the corresponding manufacturer of the COVID-19 vaccine administered as the first dose by selecting a value from the pick list	Fournissez le nom et le fabricant correspondant du vaccin contre la COVID-19 administré comme première dose en sélectionnant une valeur à partir de la liste de sélection.	Pfizer-BioNTech (Comirnaty)	Pfizer-BioNTech (Comirnaty)			PH_VACCINATION_HISTORY		
 	Host vaccination information	Informations sur la vaccination de l'hôte	GENEPIO:0100314	vaccination dose 1 vaccination date	vaccination_dose_1_vaccination_date	Dose du vaccin 1 – Date du vaccin	date	NullValueMenu						{today}			The date the first dose of a vaccine was administered.	Date à laquelle la première dose d’un vaccin a été administrée	Provide the date the first dose of COVID-19 vaccine was administered. The date should be provided in ISO 8601 standard format "YYYY-MM-DD".	Indiquez la date à laquelle la première dose du vaccin contre la COVID-19 a été administrée. La date doit être fournie selon le format standard ISO 8601 « AAAA-MM-DD ».	2021-03-01	2021-03-01			PH_VACCINATION_HISTORY		

web/templates/mpox/schema.json

@@ -9909,7 +9909,6 @@
         "MpoxInternational"
       ],
       "range": "WhitespaceMinimizedString",
-      "pattern": "{UPPER_CASE}",
       "structured_pattern": {
         "syntax": "{UPPER_CASE}",
         "interpolated": true,
@@ -9939,7 +9938,6 @@
         "MpoxInternational"
       ],
       "range": "WhitespaceMinimizedString",
-      "pattern": "{UPPER_CASE}",
       "structured_pattern": {
         "syntax": "{UPPER_CASE}",
         "interpolated": true,
@@ -10009,7 +10007,7 @@
       ],
       "examples": [
         {
-          "value": "BC Centre for Disease Control"
+          "value": "BCCDC Public Health Laboratory"
         }
       ],
       "from_schema": "https://example.com/mpox",
@@ -10535,7 +10533,7 @@
     },
     "collection_method": {
       "name": "collection_method",
-      "description": "The process used to collect the sample e.g. phlebotamy, necropsy.",
+      "description": "The process used to collect the sample e.g. phlebotomy, necropsy.",
       "title": "collection method",
       "comments": [
         "Provide a descriptor if a collection method was used for sampling. Use the picklist provided in the template.  If a desired term is missing from the picklist, contact [email protected]. If not applicable, do not leave blank. Choose a null value."
@@ -10907,12 +10905,7 @@
       "domain_of": [
         "Mpox",
         "MpoxInternational"
-      ],
-      "structured_pattern": {
-        "syntax": "{Title_Case}",
-        "interpolated": true,
-        "partial_match": false
-      }
+      ]
     },
     "host_residence_geo_loc_name_country": {
       "name": "host_residence_geo_loc_name_country",
@@ -15091,7 +15084,6 @@
           ],
           "slot_group": "Database Identifiers",
           "range": "WhitespaceMinimizedString",
-          "pattern": "{UPPER_CASE}",
           "structured_pattern": {
             "syntax": "{UPPER_CASE}",
             "interpolated": true,
@@ -15125,7 +15117,6 @@
           ],
           "slot_group": "Database Identifiers",
           "range": "WhitespaceMinimizedString",
-          "pattern": "{UPPER_CASE}",
           "structured_pattern": {
             "syntax": "{UPPER_CASE}",
             "interpolated": true,
@@ -15176,7 +15167,7 @@
           ],
           "examples": [
             {
-              "value": "BC Centre for Disease Control"
+              "value": "BCCDC Public Health Laboratory"
             }
           ],
           "from_schema": "https://example.com/mpox",
@@ -15839,7 +15830,7 @@
         },
         "collection_method": {
           "name": "collection_method",
-          "description": "The process used to collect the sample e.g. phlebotamy, necropsy.",
+          "description": "The process used to collect the sample e.g. phlebotomy, necropsy.",
           "title": "collection method",
           "comments": [
             "Provide a descriptor if a collection method was used for sampling. Use the picklist provided in the template.  If a desired term is missing from the picklist, contact [email protected]. If not applicable, do not leave blank. Choose a null value."
@@ -16350,11 +16341,6 @@
           ],
           "slot_group": "Host Information",
           "required": true,
-          "structured_pattern": {
-            "syntax": "{Title_Case}",
-            "interpolated": true,
-            "partial_match": false
-          },
           "any_of": [
             {
               "range": "HostGenderMenu"
@@ -20233,7 +20219,6 @@
           ],
           "slot_group": "Database Identifiers",
           "range": "WhitespaceMinimizedString",
-          "pattern": "{UPPER_CASE}",
           "structured_pattern": {
             "syntax": "{UPPER_CASE}",
             "interpolated": true,
@@ -20267,7 +20252,6 @@
           ],
           "slot_group": "Database Identifiers",
           "range": "WhitespaceMinimizedString",
-          "pattern": "{UPPER_CASE}",
           "structured_pattern": {
             "syntax": "{UPPER_CASE}",
             "interpolated": true,
@@ -20345,7 +20329,7 @@
           ],
           "examples": [
             {
-              "value": "BC Centre for Disease Control"
+              "value": "BCCDC Public Health Laboratory"
             }
           ],
           "from_schema": "https://example.com/mpox",
@@ -20990,7 +20974,7 @@
         },
         "collection_method": {
           "name": "collection_method",
-          "description": "The process used to collect the sample e.g. phlebotamy, necropsy.",
+          "description": "The process used to collect the sample e.g. phlebotomy, necropsy.",
           "title": "collection method",
           "comments": [
             "Provide a descriptor if a collection method was used for sampling. Use the picklist provided in the template.  If a desired term is missing from the picklist, contact [email protected]. If not applicable, do not leave blank. Choose a null value."
@@ -21611,11 +21595,6 @@
           ],
           "slot_group": "Host Information",
           "recommended": true,
-          "structured_pattern": {
-            "syntax": "{Title_Case}",
-            "interpolated": true,
-            "partial_match": false
-          },
           "any_of": [
             {
               "range": "HostGenderInternationalMenu"

web/templates/mpox/schema.yaml

@@ -1753,7 +1753,6 @@ slots:
       Sequence Archive (GSA) accessions start with CRR.
     slot_uri: GENEPIO:0101203
     range: WhitespaceMinimizedString
-    pattern: '{UPPER_CASE}'
     structured_pattern:
       syntax: '{UPPER_CASE}'
       partial_match: false
@@ -1773,7 +1772,6 @@ slots:
       the GenBank accession version.
     slot_uri: GENEPIO:0101204
     range: WhitespaceMinimizedString
-    pattern: '{UPPER_CASE}'
     structured_pattern:
       syntax: '{UPPER_CASE}'
       partial_match: false
@@ -1828,7 +1826,7 @@ slots:
     slot_uri: GENEPIO:0001153
     required: true
     examples:
-    - value: BC Centre for Disease Control
+    - value: BCCDC Public Health Laboratory
   sample_collector_contact_email:
     name: sample_collector_contact_email
     title: sample collector contact email
@@ -2155,7 +2153,7 @@ slots:
   collection_method:
     name: collection_method
     title: collection method
-    description: The process used to collect the sample e.g. phlebotamy, necropsy.
+    description: The process used to collect the sample e.g. phlebotomy, necropsy.
     comments: Provide a descriptor if a collection method was used for sampling. Use
       the picklist provided in the template.  If a desired term is missing from the
       picklist, contact [email protected]. If not applicable, do not leave blank.
@@ -2355,10 +2353,6 @@ slots:
     description: The gender of the host at the time of sample collection.
     comments: If known, select a value from the pick list.
     slot_uri: GENEPIO:0001395
-    structured_pattern:
-      syntax: '{Title_Case}'
-      partial_match: false
-      interpolated: true
   host_residence_geo_loc_name_country:
     name: host_residence_geo_loc_name_country
     title: host residence geo_loc name (country)