Correct Summary population and deterministic reporting

modkit-core/src/modbam_util/subcommands.rs

@@ -1771,6 +1771,10 @@ pub struct ModSummarize {
     interval_size: u32,
 }
 
+fn sampling_fraction_log_message(fraction: f64) -> String {
+    format!("sampling {}% of reads", fraction * 100f64)
+}
+
 impl ModSummarize {
     pub fn run(&self) -> anyhow::Result<()> {
         let _handle = init_logging(self.log_filepath.as_ref());
@@ -1807,7 +1811,9 @@ impl ModSummarize {
             (None, Some(num_reads)) => {
                 info!("sampling {num_reads} reads from BAM")
             }
-            (Some(pct), None) => info!("sampling {pct}% of reads"),
+            (Some(fraction), None) => {
+                info!("{}", sampling_fraction_log_message(fraction))
+            }
             (Some(_), Some(_)) => unreachable!(),
         });
 
@@ -1932,7 +1938,11 @@ impl ModSummarize {
                 multi_progress.clone(),
             )?;
 
-            if self.mapped_only || self.matched_only || self.motif.is_some() {
+            if self.mapped_only
+                || self.matched_only
+                || self.motif.is_some()
+                || self.cpg
+            {
                 qual_hist
             } else if let Some(nr) = self.num_reads {
                 if qual_hist.ok_records < nr {
@@ -2473,3 +2483,13 @@ impl ModMode {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::sampling_fraction_log_message;
+
+    #[test]
+    fn summary_sampling_fraction_log_uses_percentage_units() {
+        assert_eq!(sampling_fraction_log_message(0.1), "sampling 10% of reads");
+    }
+}

modkit-core/src/summarize.rs

@@ -48,8 +48,13 @@ pub struct ModSummary<'a> {
 
 impl<'a> ModSummary<'a> {
     pub(crate) fn mod_bases(&self) -> String {
-        self.mod_call_counts
-            .keys()
+        [DnaBase::A, DnaBase::C, DnaBase::G, DnaBase::T]
+            .into_iter()
+            .filter(|base| {
+                self.mod_call_counts.contains_key(base)
+                    || self.filtered_mod_call_counts.contains_key(base)
+                    || self.per_base_mod_codes.contains_key(base)
+            })
             .map(|d| d.char().to_string())
             .collect::<Vec<String>>()
             .join(",")