refactoring QcReport

Author: pnrobinson

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ga4ghphetools"
-version = "0.5.4"
+version = "0.5.5"
 edition = "2021"
 keywords = ["GA4GH", "Phenopacket Schema", "Human Phenotype Ontology"]
 description = "Generate GA4GH phenopackets from tabular data"

src/repo/cohort_qc.rs

@@ -72,6 +72,12 @@ impl CohortQc {
                 .filter_map(|dqc| dqc.check_all_rows_output_as_ppkt()),
         );
 
+        errs.extend(
+            self.disease_qc_list
+            .iter()
+            .flat_map(|dqc|dqc.check_no_hpo())
+        );
+
         errs
     }
 

src/repo/dashboard_data.rs

@@ -102,4 +102,16 @@ impl DiseaseQc {
         }
     }
 
+    pub fn check_no_hpo(&self) -> Vec<QcReport> {
+        let mut errs: Vec<QcReport> = Vec::new();
+        for ppkt in &self.ppkt_list {
+            let n_hpo = ppkt.phenotypic_features.iter().filter(|p| ! p.excluded )
+            .count();
+            if n_hpo == 0 {
+                errs.push(QcReport::no_hpo(&self.disease_data_display(), &ppkt.id));
+            }
+        }
+        errs
+    }
+
 }

src/repo/disease_qc.rs

@@ -10,7 +10,6 @@ use crate::repo::{gpt_repository::GptRepository, repo_qc::RepoQc};
 
 mod cohort_dir;
 mod cohort_qc;
-pub mod dashboard_data;
 mod disease_qc;
 mod gpt_repository;
 pub mod qc_report;

src/repo/mod.rs

@@ -1,13 +1,25 @@
 use std::collections::HashSet;
 
+use serde::de::Unexpected;
+
+
+
+#[derive(Debug, Clone, serde::Serialize)] 
+#[serde(rename_all = "camelCase")]
+enum RepoErrorType {
+    UnexpectedFile,
+    MoiMismatch,
+    PpktExportError,
+    NoHpoTermError
+}
 
 
 #[derive(Debug, Clone, serde::Serialize)] 
 #[serde(rename_all = "camelCase")]
 pub struct QcReport {
     pub cohort_name: String,
     pub message: String,
-    pub is_ok: bool,
+    pub error_type: RepoErrorType,
 }
 
 
@@ -20,7 +32,7 @@ impl QcReport {
         let msg = format!("Unexpected file: {}", unexpected);
         Self { cohort_name: cohort_name.to_string(), 
             message: msg, 
-            is_ok: false 
+            error_type: RepoErrorType::UnexpectedFile 
         }
     }
 
@@ -36,7 +48,7 @@ impl QcReport {
         let message= format!("Expected counts of {} but got {} for {}.", set,ac, ppkt_id);
         Self { cohort_name: cohort_name.to_string(), 
             message, 
-            is_ok: false 
+            error_type: RepoErrorType::MoiMismatch 
         }
     }
 
@@ -45,7 +57,16 @@ impl QcReport {
         Self {
             cohort_name: cohort_name.to_string(),
             message,
-            is_ok: false,
+            error_type: RepoErrorType::PpktExportError,
+        }
+    }
+
+    pub fn no_hpo(cohort_name: &str, ppkt_id: &str) -> Self {
+        let message = format!("Phenopacket {} had no observed HPO terms", ppkt_id);
+        Self {
+            cohort_name: cohort_name.to_string(),
+            message,
+            error_type: RepoErrorType::NoHpoTermError
         }
     }
 

src/repo/qc_report.rs

@@ -105,25 +105,6 @@ pub fn acvr1_cohort_with_repeated_term(
 }
 
 
-#[rstest]
-fn export_with_het(
-    hpo: Arc<FullCsrOntology>
-) {
-    let cohort = "/Users/robin/GIT/phenopacket-store/notebooks/KLF1/KLF1_CDAN4B_individuals.json";
-    let file_data = std::fs::read_to_string(cohort)
-            .map_err(|e| 
-                format!("Could not extract string data from {}: {}", cohort, e.to_string())).unwrap();
-        let cohort: CohortData = serde_json::from_str(&file_data)
-            .map_err(|e| format!("Could not transform string {} to CohortDto: {}",
-                file_data, e.to_string())).unwrap();
-    let orcid = "0000-0000-0000-0001".to_string();
-    let ppkt_list = ga4ghphetools::ppkt::get_phenopackets(cohort, orcid, hpo).unwrap();
-    let interp = ppkt_list[2].interpretations[0].clone();
-    if let Some(dx) = interp.diagnosis {
-        println!("# genomic interpretations = {}", dx.genomic_interpretations.len());
-    }
-    
-}