@sjcrh/proteinpaint-rust 2.74.0 → 2.75.0

package/package.json

@@ -1,5 +1,5 @@
 {
-	"version": "2.74.0",
+	"version": "2.75.0",
 	"name": "@sjcrh/proteinpaint-rust",
 	"description": "Rust-based utilities for proteinpaint",
 	"main": "index.js",
@@ -38,5 +38,5 @@
 	"devDependencies": {
 		"tape": "^5.2.2"
 	},
-	"pp_release_tag": "v2.74.0"
+	"pp_release_tag": "v2.75.0"
 }

package/src/genesetORA.rs

@@ -34,23 +34,37 @@ struct pathway_p_value {
     pathway_name: String,
     p_value_original: f64,
     p_value_adjusted: Option<f64>,
+    gene_set_hits: String,
+    gene_set_size: usize,
 }
 
 fn calculate_hypergeometric_p_value(
     sample_genes: &Vec<&str>,
     num_background_genes: usize,
     genes_in_pathway: Vec<pathway_genes>,
-) -> f64 {
-    let matching_sample_genes_counts: f64 = sample_genes
-        .iter()
-        .zip(&genes_in_pathway)
-        .filter(|&(a, b)| *a.to_string() == b.symbol)
-        .count() as f64;
+) -> (f64, f64, String) {
+    let mut matching_sample_genes_counts = 0.0;
+    let mut gene_set_hits: String = "".to_string();
+    for gene in sample_genes {
+        for pathway in &genes_in_pathway {
+            if pathway.symbol == gene.to_string() {
+                matching_sample_genes_counts += 1.0;
+                gene_set_hits += &(gene.to_string() + &",");
+            }
+        }
+    }
+
+    if matching_sample_genes_counts > 0.0 {
+        gene_set_hits.pop();
+    }
+
+    //println!("sample_genes:{:?}", sample_genes);
+    //println!("genes_in_pathway:{:?}", genes_in_pathway);
     //println!("k-1:{}", matching_sample_genes_counts - 1.0);
     //println!("M:{}", genes_in_pathway.len() as f64);
     //println!(
     //    "N-M:{}",
-    //    background_genes.len() as f64 - genes_in_pathway.len() as f64
+    //    num_background_genes as f64 - genes_in_pathway.len() as f64
     //);
     //println!("n:{}", sample_genes.len() as f64);
     let p_value = r_mathlib::hypergeometric_cdf(
@@ -62,7 +76,7 @@ fn calculate_hypergeometric_p_value(
         false,
     );
     //println!("p_value:{}", p_value);
-    p_value
+    (p_value, matching_sample_genes_counts, gene_set_hits)
 }
 
 fn main() -> Result<()> {
@@ -136,7 +150,6 @@ fn main() -> Result<()> {
                             + &genesetgroup
                             + "'"),
                     );
-                    let mut iter = 0;
                     match stmt_result {
                         Ok(mut stmt) => {
                             #[allow(non_snake_case)]
@@ -144,7 +157,6 @@ fn main() -> Result<()> {
                                 stmt.query_map([], |row| Ok(GO_pathway { GO_id: row.get(0)? }))?;
                             #[allow(non_snake_case)]
                             for GO_term in GO_iter {
-                                iter += 1;
                                 match GO_term {
                                     Ok(n) => {
                                         //println!("GO term {:?}", n);
@@ -184,16 +196,20 @@ fn main() -> Result<()> {
                                                 }
                                             }
                                         }
-                                        let p_value = calculate_hypergeometric_p_value(
-                                            &sample_genes,
-                                            num_background_genes,
-                                            names,
-                                        );
-                                        if p_value.is_nan() == false {
+                                        let gene_set_size = names.len();
+                                        let (p_value, matches, gene_set_hits) =
+                                            calculate_hypergeometric_p_value(
+                                                &sample_genes,
+                                                num_background_genes,
+                                                names,
+                                            );
+                                        if matches >= 1.0 && p_value.is_nan() == false {
                                             pathway_p_values.push(pathway_p_value {
                                                 pathway_name: n.GO_id,
                                                 p_value_original: p_value,
                                                 p_value_adjusted: None,
+                                                gene_set_hits: gene_set_hits,
+                                                gene_set_size: gene_set_size,
                                             })
                                         }
                                     }
@@ -206,7 +222,7 @@ fn main() -> Result<()> {
                         Err(_) => panic!("sqlite database file not found"),
                     }
                     let output_string = "{\"num_pathways\":".to_string()
-                        + &iter.to_string()
+                        + &pathway_p_values.len().to_string()
                         + &",\"pathways\":"
                         + &adjust_p_values(pathway_p_values, num_items_output)
                         + &"}";
@@ -263,6 +279,8 @@ fn adjust_p_values(
             pathway_name: original_p_values[i].pathway_name.clone(),
             p_value_original: original_p_values[i].p_value_original,
             p_value_adjusted: Some(adjusted_p_val),
+            gene_set_hits: original_p_values[i].gene_set_hits.clone(),
+            gene_set_size: original_p_values[i].gene_set_size,
         });
     }
     adjusted_p_values.as_mut_slice().sort_by(|a, b| {
@@ -277,8 +295,7 @@ fn adjust_p_values(
 
     let mut output_string = "[".to_string();
     for i in 0..num_items_output {
-        let j = adjusted_p_values.len() - i - 1;
-        output_string += &serde_json::to_string(&adjusted_p_values[j]).unwrap();
+        output_string += &serde_json::to_string(&adjusted_p_values[i]).unwrap();
         if i < num_items_output - 1 {
             output_string += &",".to_string();
         }