Fix Int overflow for bases_in_reads on large HiFi samples

read_int uses Integer.parseInt which caps at ~2.1 Gbp; samples with
>2.1 Gbp of reads (e.g. 9.5 Gbp) caused t_02_HifiSeqkitStats to fail.
Change bases_in_reads, bases_in_reads_over_10kb, bases_in_reads_over_20kb
to Float/read_float throughout. Wrap bash interpolation in printf "%.0f"
so $(( )) integer arithmetic handles Cromwell's scientific-notation output.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: tpshea2

wdl/pipelines/PacBio/Metagenomics/HvpReadProcessing.wdl

@@ -114,7 +114,7 @@ workflow HvpReadProcessing {
         File  seqkit_stats_tsv            = t_02_HifiSeqkitStats.seqkit_stats_tsv
         File  seqkit_fx2tab_tsv           = t_02_HifiSeqkitStats.seqkit_fx2tab_tsv
         Int   num_reads                   = t_02_HifiSeqkitStats.num_reads
-        Int   bases_in_reads              = t_02_HifiSeqkitStats.bases_in_reads
+        Float bases_in_reads              = t_02_HifiSeqkitStats.bases_in_reads
         Int   max_read_length             = t_02_HifiSeqkitStats.max_read_length
         Int   q1_read_length              = t_02_HifiSeqkitStats.q1_read_length
         Int   median_read_length          = t_02_HifiSeqkitStats.median_read_length
@@ -123,8 +123,8 @@ workflow HvpReadProcessing {
         Float pct_q20_bases               = t_02_HifiSeqkitStats.pct_q20_bases
         Float pct_q30_bases               = t_02_HifiSeqkitStats.pct_q30_bases
         Float mean_read_gc                = t_02_HifiSeqkitStats.mean_read_gc
-        Int   bases_in_reads_over_10kb    = t_02_HifiSeqkitStats.bases_in_reads_over_10kb
-        Int   bases_in_reads_over_20kb    = t_02_HifiSeqkitStats.bases_in_reads_over_20kb
+        Float bases_in_reads_over_10kb    = t_02_HifiSeqkitStats.bases_in_reads_over_10kb
+        Float bases_in_reads_over_20kb    = t_02_HifiSeqkitStats.bases_in_reads_over_20kb
 
         File  spikein_report        = t_03_SpikeInRemoval.spikein_report
         File  spikein_stats_tsv     = t_03_SpikeInRemoval.spikein_stats_tsv

wdl/pipelines/PacBio/QC/HifiReadQCPipeline.wdl

@@ -134,11 +134,11 @@ workflow HifiReadQCPipeline {
         String expected_genome_size     = t_01_GetTaxIdAndGenomeSize.expected_genome_size
 
         Int    num_reads                = t_03_HifiSeqkitStats.num_reads
-        Int    bases_in_reads           = t_03_HifiSeqkitStats.bases_in_reads
+        Float  bases_in_reads           = t_03_HifiSeqkitStats.bases_in_reads
         String estimate_cvg             = t_05_CreateHifiQCReport.estimate_cvg
-        Int    bases_in_reads_over_10kb = t_03_HifiSeqkitStats.bases_in_reads_over_10kb
+        Float  bases_in_reads_over_10kb = t_03_HifiSeqkitStats.bases_in_reads_over_10kb
         String estimate_cvg_reads_10kb  = t_05_CreateHifiQCReport.estimate_cvg_reads_10kb
-        Int    bases_in_reads_over_20kb = t_03_HifiSeqkitStats.bases_in_reads_over_20kb
+        Float  bases_in_reads_over_20kb = t_03_HifiSeqkitStats.bases_in_reads_over_20kb
         String estimate_cvg_reads_20kb  = t_05_CreateHifiQCReport.estimate_cvg_reads_20kb
         Int    q1_read_length           = t_03_HifiSeqkitStats.q1_read_length
         Int    median_read_length       = t_03_HifiSeqkitStats.median_read_length

wdl/tasks/QC/HifiQC.wdl

@@ -121,7 +121,7 @@ task HifiSeqkitStats {
         File seqkit_fx2tab_tsv        = "~{fq_basename}.seqkit_fx2tab.tsv"
 
         Int   num_reads                = read_int("stat.num_reads.txt")
-        Int   bases_in_reads           = read_int("stat.bases_in_reads.txt")
+        Float bases_in_reads           = read_float("stat.bases_in_reads.txt")
         Int   max_read_length          = read_int("stat.max_read_length.txt")
         Int   q1_read_length           = read_int("stat.q1_read_length.txt")
         Int   median_read_length       = read_int("stat.median_read_length.txt")
@@ -130,8 +130,8 @@ task HifiSeqkitStats {
         Float pct_q20_bases            = read_float("stat.pct_q20_bases.txt")
         Float pct_q30_bases            = read_float("stat.pct_q30_bases.txt")
         Float mean_read_gc             = read_float("stat.mean_read_gc.txt")
-        Int   bases_in_reads_over_10kb = read_int("stat.bases_in_reads_over_10kb.txt")
-        Int   bases_in_reads_over_20kb = read_int("stat.bases_in_reads_over_20kb.txt")
+        Float bases_in_reads_over_10kb = read_float("stat.bases_in_reads_over_10kb.txt")
+        Float bases_in_reads_over_20kb = read_float("stat.bases_in_reads_over_20kb.txt")
     }
 
     #########################
@@ -388,9 +388,9 @@ task CreateHifiQCReport {
         String expected_genome_size
 
         Int    num_reads
-        Int    bases_in_reads
-        Int    bases_in_reads_over_10kb
-        Int    bases_in_reads_over_20kb
+        Float  bases_in_reads
+        Float  bases_in_reads_over_10kb
+        Float  bases_in_reads_over_20kb
         Int    q1_read_length
         Int    median_read_length
         Int    q3_read_length
@@ -438,9 +438,9 @@ task CreateHifiQCReport {
         # ---- end preamble ----
 
         EXPECTED_GENOME_SIZE="~{expected_genome_size}"
-        SUM_LEN=~{bases_in_reads}
-        B10K=~{bases_in_reads_over_10kb}
-        B20K=~{bases_in_reads_over_20kb}
+        SUM_LEN=$(printf "%.0f" ~{bases_in_reads})
+        B10K=$(printf "%.0f" ~{bases_in_reads_over_10kb})
+        B20K=$(printf "%.0f" ~{bases_in_reads_over_20kb})
 
         if [[ "${EXPECTED_GENOME_SIZE}" =~ ^[0-9]+$ ]] && [[ "${EXPECTED_GENOME_SIZE}" -gt 0 ]]; then
             EST_CVG=$(( SUM_LEN / EXPECTED_GENOME_SIZE ))