fix: regenerate .fai index before reading and guard empty edx samples

Regenerate the FASTA index in compute_seq_similarity.py to prevent
stale .fai from referencing collapsed-away sequences. Also guard
edx_concordance output on actual edx samples existing, not just
edx.enabled config flag.

Author: jayhesselberth

workflow/rules/common.smk

@@ -315,7 +315,7 @@ def pipeline_outputs():
     outs.append(os.path.join(outdir, "summary", "qc", "reference_similarity.tsv"))
 
     # EDX (3' adapter barcode) concordance table
-    if config.get("edx", {}).get("enabled", False):
+    if config.get("edx", {}).get("enabled", False) and get_edx_samples():
         outs.append(os.path.join(outdir, "summary", "edx", "edx_concordance.tsv.gz"))
 
     return outs

workflow/scripts/compute_seq_similarity.py

@@ -32,6 +32,9 @@ def compute_similarity_matrix(fasta_path):
         - similarity_matrix: numpy array of percent identity values
         - sequence_names: list of sequence IDs
     """
+    # Regenerate .fai index to ensure it matches the current FASTA content
+    pysam.faidx(fasta_path)
+
     # Read all sequences from FASTA using pysam
     faidx = pysam.FastaFile(fasta_path)
     names = list(faidx.references)