fix: ensure microsatellite indicator is an f32

Author: clintval

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "vartovcf"
-version = "1.3.0"
+version = "1.4.0"
 authors = ["Clint Valentine <valentine.clint@gmail.com>"]
 edition = "2024"
 license = "MIT"

src/lib/record.rs

@@ -289,7 +289,7 @@ pub struct TumorOnlyVariant<'a> {
     /// The number of bases to be shifted 3-prime for deletions due to alternative alignment(s).
     pub num_bases_3_prime_shift_for_deletions: i32,
     /// Whether the variant call is in a microsatellite (MSI) or not. Greater than 1 indicates MSI.
-    pub microsatellite: i32,
+    pub microsatellite: f32,
     /// The length of the microsatellite in base pairs of reference genome.
     pub microsatellite_length: i32,
     /// The length of the microsatellite in base pairs of reference genome.
@@ -432,12 +432,12 @@ mod tests {
         let dup_info = SvInfo { supporting_split_reads: 1, supporting_pairs: 1, supporting_clusters: 1 };
         let del_info = SvInfo { supporting_split_reads: 1, supporting_pairs: 1, supporting_clusters: 1 };
         vec!(
-            TumorOnlyVariant { sample: "dna00001", interval_name: "PTPN11", contig: "chr12", start: 112450447, end: 123513818, ref_allele: "A", alt_allele: "<INV>", depth: 6775, alt_depth: 1, ref_forward: 3991, ref_reverse: 2588, alt_forward: 1, alt_reverse: 0, gt: "A/<INV11063372>", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 58.0, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 33.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0002, af_adjusted: 0.0001, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 1, high_quality_total_reads: 6582, flank_seq_5_prime: "GAACATCACGGGCAATTAAA", flank_seq_3_prime: "GGGACCTAGATTTTAAGAGA", segment: "chr12:112450168-112450587", variant_type: "INV", duplication_rate: None, sv_info: Some(inv_info), distance_to_crispr_site: None },
-            TumorOnlyVariant { sample: "dna00001", interval_name: "PTPN11", contig: "chr12", start: 112450447, end: 123513818, ref_allele: "A", alt_allele: "<DUP>", depth: 6775, alt_depth: 1, ref_forward: 3991, ref_reverse: 2588, alt_forward: 1, alt_reverse: 0, gt: "A/<DUP11063372>", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 58.0, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 10.0, mean_mapping_quality: 33.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0002, af_adjusted: 0.0001, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 1, high_quality_total_reads: 6582, flank_seq_5_prime: "GAACATCACGGGCAATTAAA", flank_seq_3_prime: "GGGACCTAGATTTTAAGAGA", segment: "chr12:112450168-112450587", variant_type: "DUP", duplication_rate: Some(0.001), sv_info: Some(dup_info), distance_to_crispr_site: None },
-            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713883, end: 114713883, ref_allele: "G", alt_allele: "A", depth: 8104, alt_depth: 1, ref_forward: 2766, ref_reverse: 5280, alt_forward: 1, alt_reverse: 0, gt: "G/A", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 13.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 0.34385, strand_bias_odds_ratio: 4.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 1, microsatellite_length: 1, mean_mismatches_in_reads: 2.0, high_quality_variant_reads: 1, high_quality_total_reads: 8048, flank_seq_5_prime: "TCGCCTGTCCTCATGTATTG", flank_seq_3_prime: "TCTCTCATGGCACTGTACTC", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
-            TumorOnlyVariant { sample: "dna00001", interval_name: "FLT3", contig: "chr13", start: 24684729, end: 28034141, ref_allele: "G", alt_allele: "<DEL>", depth: 7463, alt_depth: 2, ref_forward: 7463, ref_reverse: 0, alt_forward: 0, alt_reverse: 2, gt: "-3349412/-3349412", af: 0.0003, strand_bias: PairBias::from_str("0;0").unwrap(), mean_position_in_read: 60.5, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 2.0, mean_mapping_quality: 60.0, signal_to_noise: 4.0, af_high_quality_bases: 1.0000, af_adjusted: 0.0003, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 2, high_quality_total_reads: 2, flank_seq_5_prime: "TGCTGTAGTCTAATGATTCT", flank_seq_3_prime: "CAACGTAGAAGTACTCATTA", segment: "chr13:28033879-28034298", variant_type: "DEL", duplication_rate: None, sv_info: Some(del_info), distance_to_crispr_site: None },
-            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713883, end: 114713883, ref_allele: "G", alt_allele: "T", depth: 8104, alt_depth: 1, ref_forward: 2766, ref_reverse: 5280, alt_forward: 0, alt_reverse: 1, gt: "G/T", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 28.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 1, microsatellite_length: 1, mean_mismatches_in_reads: 1.0, high_quality_variant_reads: 1, high_quality_total_reads: 8048, flank_seq_5_prime: "TCGCCTGTCCTCATGTATTG", flank_seq_3_prime: "TCTCTCATGGCACTGTACTC", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
-            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713880, end: 114713880, ref_allele: "T", alt_allele: "A", depth: 8211, alt_depth: 1, ref_forward: 3001, ref_reverse: 5130, alt_forward: 1, alt_reverse: 0, gt: "T/A", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 18.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 0.36916, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 2, microsatellite_length: 1, mean_mismatches_in_reads: 1.0, high_quality_variant_reads: 1, high_quality_total_reads: 8132, flank_seq_5_prime: "CCTTCGCCTGTCCTCATGTA", flank_seq_3_prime: "TGGTCTCTCATGGCACTGTA", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "PTPN11", contig: "chr12", start: 112450447, end: 123513818, ref_allele: "A", alt_allele: "<INV>", depth: 6775, alt_depth: 1, ref_forward: 3991, ref_reverse: 2588, alt_forward: 1, alt_reverse: 0, gt: "A/<INV11063372>", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 58.0, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 33.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0002, af_adjusted: 0.0001, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0.0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 1, high_quality_total_reads: 6582, flank_seq_5_prime: "GAACATCACGGGCAATTAAA", flank_seq_3_prime: "GGGACCTAGATTTTAAGAGA", segment: "chr12:112450168-112450587", variant_type: "INV", duplication_rate: None, sv_info: Some(inv_info), distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "PTPN11", contig: "chr12", start: 112450447, end: 123513818, ref_allele: "A", alt_allele: "<DUP>", depth: 6775, alt_depth: 1, ref_forward: 3991, ref_reverse: 2588, alt_forward: 1, alt_reverse: 0, gt: "A/<DUP11063372>", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 58.0, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 10.0, mean_mapping_quality: 33.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0002, af_adjusted: 0.0001, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0.0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 1, high_quality_total_reads: 6582, flank_seq_5_prime: "GAACATCACGGGCAATTAAA", flank_seq_3_prime: "GGGACCTAGATTTTAAGAGA", segment: "chr12:112450168-112450587", variant_type: "DUP", duplication_rate: Some(0.001), sv_info: Some(dup_info), distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713883, end: 114713883, ref_allele: "G", alt_allele: "A", depth: 8104, alt_depth: 1, ref_forward: 2766, ref_reverse: 5280, alt_forward: 1, alt_reverse: 0, gt: "G/A", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 13.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 0.34385, strand_bias_odds_ratio: 4.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 1.0, microsatellite_length: 1, mean_mismatches_in_reads: 2.0, high_quality_variant_reads: 1, high_quality_total_reads: 8048, flank_seq_5_prime: "TCGCCTGTCCTCATGTATTG", flank_seq_3_prime: "TCTCTCATGGCACTGTACTC", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "FLT3", contig: "chr13", start: 24684729, end: 28034141, ref_allele: "G", alt_allele: "<DEL>", depth: 7463, alt_depth: 2, ref_forward: 7463, ref_reverse: 0, alt_forward: 0, alt_reverse: 2, gt: "-3349412/-3349412", af: 0.0003, strand_bias: PairBias::from_str("0;0").unwrap(), mean_position_in_read: 60.5, stdev_position_in_read: 1.0, base_quality_mean: 90.0, stdev_base_stdev: 1.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 2.0, mean_mapping_quality: 60.0, signal_to_noise: 4.0, af_high_quality_bases: 1.0000, af_adjusted: 0.0003, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 0.0, microsatellite_length: 0, mean_mismatches_in_reads: 0.0, high_quality_variant_reads: 2, high_quality_total_reads: 2, flank_seq_5_prime: "TGCTGTAGTCTAATGATTCT", flank_seq_3_prime: "CAACGTAGAAGTACTCATTA", segment: "chr13:28033879-28034298", variant_type: "DEL", duplication_rate: None, sv_info: Some(del_info), distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713883, end: 114713883, ref_allele: "G", alt_allele: "T", depth: 8104, alt_depth: 1, ref_forward: 2766, ref_reverse: 5280, alt_forward: 0, alt_reverse: 1, gt: "G/T", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 28.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 1.0, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 1.0, microsatellite_length: 1, mean_mismatches_in_reads: 1.0, high_quality_variant_reads: 1, high_quality_total_reads: 8048, flank_seq_5_prime: "TCGCCTGTCCTCATGTATTG", flank_seq_3_prime: "TCTCTCATGGCACTGTACTC", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
+            TumorOnlyVariant { sample: "dna00001", interval_name: "NRAS-Q61", contig: "chr1", start: 114713880, end: 114713880, ref_allele: "T", alt_allele: "A", depth: 8211, alt_depth: 1, ref_forward: 3001, ref_reverse: 5130, alt_forward: 1, alt_reverse: 0, gt: "T/A", af: 0.0001, strand_bias: PairBias::from_str("2;0").unwrap(), mean_position_in_read: 18.0, stdev_position_in_read: 0.0, base_quality_mean: 90.0, stdev_base_stdev: 0.0, strand_bias_p_value: 0.36916, strand_bias_odds_ratio: 0.0, mean_mapping_quality: 60.0, signal_to_noise: 2.0, af_high_quality_bases: 0.0001, af_adjusted: 0.0, num_bases_3_prime_shift_for_deletions: 0, microsatellite: 2.0, microsatellite_length: 1, mean_mismatches_in_reads: 1.0, high_quality_variant_reads: 1, high_quality_total_reads: 8132, flank_seq_5_prime: "CCTTCGCCTGTCCTCATGTA", flank_seq_3_prime: "TGGTCTCTCATGGCACTGTA", segment: "chr1:114713749-114713988", variant_type: "SNV", duplication_rate: None, sv_info: None, distance_to_crispr_site: None },
         )
     }
 

tests/calls.duplicates.vcf

@@ -1,5 +1,5 @@
 ##fileformat=VCFv4.2
-##source=vartovcf-1.3.0
+##source=vartovcf-1.4.0
 ##INFO=<ID=END,Number=1,Type=Integer,Description="The end location of this variant call.">
 ##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="The length of structural variant in base pairs of reference genome, if this call is a structural variant.">
 ##INFO=<ID=SVTYPE,Number=1,Type=String,Description="The structural variant type (BND, CNV, DEL, DUP, INS, INV), if this call is a structural variant.">

tests/calls.g.vcf

@@ -1,5 +1,5 @@
 ##fileformat=VCFv4.2
-##source=vartovcf-1.3.0
+##source=vartovcf-1.4.0
 ##INFO=<ID=END,Number=1,Type=Integer,Description="The end location of this variant call.">
 ##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="The length of structural variant in base pairs of reference genome, if this call is a structural variant.">
 ##INFO=<ID=SVTYPE,Number=1,Type=String,Description="The structural variant type (BND, CNV, DEL, DUP, INS, INV), if this call is a structural variant.">

tests/calls.non-variants-skipped.vcf

@@ -1,5 +1,5 @@
 ##fileformat=VCFv4.2
-##source=vartovcf-1.3.0
+##source=vartovcf-1.4.0
 ##INFO=<ID=END,Number=1,Type=Integer,Description="The end location of this variant call.">
 ##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="The length of structural variant in base pairs of reference genome, if this call is a structural variant.">
 ##INFO=<ID=SVTYPE,Number=1,Type=String,Description="The structural variant type (BND, CNV, DEL, DUP, INS, INV), if this call is a structural variant.">

tests/calls.skippable.vcf

@@ -1,5 +1,5 @@
 ##fileformat=VCFv4.2
-##source=vartovcf-1.3.0
+##source=vartovcf-1.4.0
 ##INFO=<ID=END,Number=1,Type=Integer,Description="The end location of this variant call.">
 ##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="The length of structural variant in base pairs of reference genome, if this call is a structural variant.">
 ##INFO=<ID=SVTYPE,Number=1,Type=String,Description="The structural variant type (BND, CNV, DEL, DUP, INS, INV), if this call is a structural variant.">

tests/calls.vcf

@@ -1,5 +1,5 @@
 ##fileformat=VCFv4.2
-##source=vartovcf-1.3.0
+##source=vartovcf-1.4.0
 ##INFO=<ID=END,Number=1,Type=Integer,Description="The end location of this variant call.">
 ##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="The length of structural variant in base pairs of reference genome, if this call is a structural variant.">
 ##INFO=<ID=SVTYPE,Number=1,Type=String,Description="The structural variant type (BND, CNV, DEL, DUP, INS, INV), if this call is a structural variant.">