From 9690f12e6725447e035903d3b4eecf31d01b1576 Mon Sep 17 00:00:00 2001
From: Thomas Battenfeld <46334240+thomasbtf@users.noreply.github.com>
Date: Sun, 19 Dec 2021 18:32:11 +0100
Subject: [PATCH] fix: pangolin input for overview table, bamclipper samtools
 error, include flag for high quality genome report (#418)

* fixes from production

* fmt
---
 workflow/envs/bamclipper.yaml      |  1 +
 workflow/rules/common.smk          | 26 +++++++++++++++++++++++++-
 workflow/rules/generate_output.smk |  4 +---
 3 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/workflow/envs/bamclipper.yaml b/workflow/envs/bamclipper.yaml
index f88e6e2e9..c0cc02571 100644
--- a/workflow/envs/bamclipper.yaml
+++ b/workflow/envs/bamclipper.yaml
@@ -3,3 +3,4 @@ channels:
   - conda-forge
 dependencies:
   - bamclipper =1.0
+  - samtools =1.9 # see https://github.com/merenlab/anvio/issues/1479
diff --git a/workflow/rules/common.smk b/workflow/rules/common.smk
index 8c08b4520..77e8f233c 100644
--- a/workflow/rules/common.smk
+++ b/workflow/rules/common.smk
@@ -1273,7 +1273,10 @@ def get_include_flag(sample):
 
 
 def get_include_flag_for_date(wildcards):
-    return [get_include_flag(sample) for sample in get_samples_for_date(wildcards.date)]
+    return [
+        get_include_flag(sample)
+        for sample in get_assemblies_for_submission(wildcards, "accepted samples")
+    ]
 
 
 def get_artic_primer(wildcards):
@@ -1541,6 +1544,27 @@ def get_aggregated_pangolin_calls(wildcards, return_list="paths"):
     return expanded_patterns
 
 
+def get_pangolin_for_report(wildcards):
+    paths = []
+
+    path = "results/{date}/tables/strain-calls/{sample}.{stage}.strains.pangolin.csv"
+
+    for entry in get_assemblies_for_submission(wildcards, "all samples"):
+        sample, assembly = entry.split(",")
+        if assembly == "normal":
+            pango_stage = "polished"
+        elif assembly == "pseudo":
+            pango_stage = "pseudo"
+        elif assembly == "consensus":
+            pango_stage = "consensus"
+        elif assembly == "not-accepted":
+            pango_stage = "polished"
+
+        paths.append(path.format(sample=sample, date=wildcards.date, stage=pango_stage))
+
+    return paths
+
+
 wildcard_constraints:
     sample="[^/.]+",
     vartype="|".join(VARTYPES),
diff --git a/workflow/rules/generate_output.smk b/workflow/rules/generate_output.smk
index e6fb8232d..26099a51e 100644
--- a/workflow/rules/generate_output.smk
+++ b/workflow/rules/generate_output.smk
@@ -134,9 +134,7 @@ rule overview_table_csv:
         pseudo_contigs=get_fallbacks_for_report("pseudo"),
         consensus_contigs=get_fallbacks_for_report("consensus"),
         kraken=get_kraken_output,
-        pangolin=expand_samples_for_date(
-            "results/{{date}}/tables/strain-calls/{sample}.polished.strains.pangolin.csv",
-        ),
+        pangolin=get_pangolin_for_report,
         bcf=expand_samples_for_date(
             "results/{{date}}/filtered-calls/ref~main/{sample}.subclonal.high+moderate-impact.bcf",
         ),