make tests pass with mc

wdecoster · May 16, 2024 · 300be24 · 300be24
1 parent f08e5b1
commit 300be24
Show file tree

Hide file tree

Showing 5 changed files with 41 additions and 13 deletions.
diff --git a/src/consensus.rs b/src/consensus.rs
@@ -198,6 +198,7 @@ mod tests {
                 start: 1,
                 end: 100,
                 id: "".to_string(),
+                motifs: "CA".to_string(),
             },
         );
         println!("Consensus: {}", cons.seq.unwrap());

diff --git a/src/genotype.rs b/src/genotype.rs
@@ -331,6 +331,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: String::from("test-repeat"),
+            motifs: "CA".to_string(),
         };
         let flanking = 2000;
         let minlen = 5;
@@ -371,6 +372,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: "TEST".to_string(),
+            motifs: "CA".to_string(),
         };
         let args = Cli {
             bam: String::from("test_data/small-test-phased.bam"),
@@ -400,6 +402,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: "TEST".to_string(),
+            motifs: "CA".to_string(),
         };
         let args = Cli {
             bam: String::from("test_data/small-test-phased.bam"),
@@ -445,6 +448,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: "TEST".to_string(),
+            motifs: "CA".to_string(),
         };
         let mut bam = parse_bam::create_bam_reader(&args.bam, &args.fasta);
         let genotype = genotype_repeat(&repeat, &args, &mut bam);
@@ -474,6 +478,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: "TEST".to_string(),
+            motifs: "CA".to_string(),
         };
         let mut bam = parse_bam::create_bam_reader(&args.bam, &args.fasta);
         let genotype = genotype_repeat(&repeat, &args, &mut bam);
@@ -504,6 +509,7 @@ mod tests {
             start: 154654404,
             end: 154654432,
             id: "TEST".to_string(),
+            motifs: "CA".to_string(),
         };
         let mut bam = parse_bam::create_bam_reader(&args.bam, &args.fasta);
         let genotype = genotype_repeat(&repeat, &args, &mut bam);

diff --git a/src/motif.rs b/src/motif.rs
@@ -6,19 +6,18 @@ fn create_motif(seq: &str) -> String {
 }
 
 pub fn create_mc(motifs_string: &str, repeats: &str) -> String {
-    let motifs: Vec<String> = motifs_string.split(',').map(|s| s.to_string()).collect();
-
-    let mut motif_counts = vec![0; motifs.len()];
-    for motif in motifs.iter() {
-        let mut start = 0;
-        while let Some(index) = repeats[start..].find(motif) {
-            motif_counts[motifs.iter().position(|x| x == motif).unwrap()] += 1;
-            start += index + motif.len();
-        }
-    }
-    motif_counts.iter().map(|count| count.to_string()).collect::<Vec<String>>().join("_")
-}
+     let motifs: Vec<String> = motifs_string.split(',').map(|s| s.to_string()).collect();
 
+     let mut motif_counts = vec![0; motifs.len()];
+     for motif in motifs.iter() {
+         let mut start = 0;
+         while let Some(index) = repeats[start..].find(motif) {
+             motif_counts[motifs.iter().position(|x| x == motif).unwrap()] += 1;
+             start += index + motif.len();
+         }
+     }
+     motif_counts.iter().map(|count| count.to_string()).collect::<Vec<String>>().join("_")
+ }
 
 #[cfg(test)]
 mod tests {
@@ -32,11 +31,22 @@ mod tests {
             "(CAG)4(CGG)3(CAG)3"
         );
     }
+    #[test]
     fn test_create_mc() {
         let motifs = "ATG,CGT,GCA";
         let repeat = "ATGCGTATGCGTAGCGT";
+        println!("{:?}", create_mc(&motifs, &repeat));
+        assert_eq!(
+            create_mc(&motifs, repeat), "2_3_0"
+        );
+    }
+    #[test]
+    fn test_create_mc_with_n() {
+        let motifs = "NCG";
+        let repeat = "ACGACGACGACG";
+
         assert_eq!(
-            create_mc(&motifs, repeat), "3_1_0"
+            create_mc(&motifs, repeat), "4"
         );
     }
 }
diff --git a/src/parse_bam.rs b/src/parse_bam.rs
@@ -144,6 +144,7 @@ fn test_get_overlapping_reads() {
         start: 154654404,
         end: 154654432,
         id: "TEST".to_string(),
+        motifs: "CA".to_string(),
     };
     let unphased = false;
     let mut bam = create_bam_reader(&bam, &fasta);
@@ -159,6 +160,7 @@ fn test_get_overlapping_reads_url1() {
         start: 154654404,
         end: 154654432,
         id: "TEST".to_string(),
+        motifs: "CA".to_string(),
     };
     let unphased = false;
     let mut bam = create_bam_reader(&bam, &fasta);
@@ -174,6 +176,7 @@ fn test_get_overlapping_reads_url2() {
         start: 154654404,
         end: 154654432,
         id: "TEST".to_string(),
+        motifs: "CA".to_string(),
     };
     let unphased = false;
     let mut bam = create_bam_reader(&bam, &fasta);
@@ -189,6 +192,7 @@ fn test_get_overlapping_reads_url3() {
         start: 154654404,
         end: 154654432,
         id: "TEST".to_string(),
+        motifs: "CA".to_string(),
     };
     let unphased = false;
     let mut bam = create_bam_reader(&bam, &fasta);
@@ -204,6 +208,7 @@ fn test_get_overlapping_reads_url4() {
         start: 154654404,
         end: 154654432,
         id: "TEST".to_string(),
+        motifs: "CA".to_string(),
     };
     let unphased = false;
     let mut bam = create_bam_reader(&bam, &fasta);

diff --git a/src/phase_insertions.rs b/src/phase_insertions.rs
@@ -333,6 +333,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );
@@ -374,6 +375,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );
@@ -414,6 +416,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );
@@ -457,6 +460,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );
@@ -498,6 +502,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST_RFC1".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );
@@ -570,6 +575,7 @@ mod tests {
                 start: 154654404,
                 end: 154654432,
                 id: "TEST_RFC1".to_string(),
+                motifs: "CA".to_string(),
             },
             false,
         );