From 179646a5d57ebae30fe99302128dc400bf391dfc Mon Sep 17 00:00:00 2001
From: Francisco Zorrilla <fz274@cam.ac.uk>
Date: Fri, 9 Jun 2023 16:40:12 +0100
Subject: [PATCH 1/3] feature: add Snakefile rule for eggnog-mapper

create Snakefile rule to annotate genomes with eggnog-mapper
---
 workflow/Snakefile | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/workflow/Snakefile b/workflow/Snakefile
index 22a9e06..9ca4add 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -1021,6 +1021,34 @@ rule binningVis:
         rm Rplots.pdf
         """
 
+rule eggnog_mapper:
+    input:
+        f'{config["path"]["root"]}/final_mags/'
+    output:
+        directory(f'{config["path"]["root"]}/emapper/')
+    benchmark:
+        f'{config["path"]["root"]}/{config["folder"]["benchmarks"]}/{{IDs}}.eggnog_mapper.benchmark.txt'
+    message:
+        """
+        batch process small number of genomes with eggnog mapper, make sure you have downloaded and 
+        configured the eggnog database path, assumes that you have ORF annotated protein fasta files 
+        (e.g. translated from DNA using prodigal) in final_mags folder, then will loop through each
+        faa file and run eggnog to produce gene annotations files, Warning: this can take a few 
+        minutes and load up to 45GB to RAM
+        """
+
+    shell:
+        """
+        # Activate eggnog environment
+        set +u;source activate eggnog6;set -u;
+
+        cd {input}
+
+        while read protein;do
+            emapper.py -m diamond -i $protein -o ${protein%.*} --cpu 18 --dbmem --excel --scratch_dir .
+        done< <(ls|grep faa)
+        """
+        
 rule abundance:
     input:
         bins = f'{config["path"]["root"]}/{config["folder"]["reassembled"]}/{{IDs}}/reassembled_bins',

From ab7e8f66ff5ff7df8ea8c29b22a553a9dd146965 Mon Sep 17 00:00:00 2001
From: Francisco Zorrilla <fz274@cam.ac.uk>
Date: Fri, 9 Jun 2023 16:48:24 +0100
Subject: [PATCH 2/3] Update Snakefile

add line to create dummy output folder, still need to replace hardcoded cores number with new param, same with eggnog env
---
 workflow/Snakefile | 1 +
 1 file changed, 1 insertion(+)

diff --git a/workflow/Snakefile b/workflow/Snakefile
index 9ca4add..42c87f6 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -1043,6 +1043,7 @@ rule eggnog_mapper:
         set +u;source activate eggnog6;set -u;
 
         cd {input}
+        mkdir -p {output}
 
         while read protein;do
             emapper.py -m diamond -i $protein -o ${protein%.*} --cpu 18 --dbmem --excel --scratch_dir .

From 725928f14fec20484e0bbb951ae7f0adf00fe6ef Mon Sep 17 00:00:00 2001
From: Francisco Zorrilla <fz274@cam.ac.uk>
Date: Fri, 9 Jun 2023 17:18:46 +0100
Subject: [PATCH 3/3] Update Snakefile eggnog_mapper rule

---
 workflow/Snakefile | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/workflow/Snakefile b/workflow/Snakefile
index 42c87f6..e36c80b 100644
--- a/workflow/Snakefile
+++ b/workflow/Snakefile
@@ -1044,9 +1044,12 @@ rule eggnog_mapper:
 
         cd {input}
         mkdir -p {output}
-
-        while read protein;do
-            emapper.py -m diamond -i $protein -o ${protein%.*} --cpu 18 --dbmem --excel --scratch_dir .
+        
+        while read protein;do             
+            mkdir -p tmp;
+            emapper.py -m diamond -i $protein -o ${protein%.*} --cpu 18 --dbmem --excel --scratch_dir tmp;
+            mv *.emapper.* {output}
+            rm -r tmp;
         done< <(ls|grep faa)
         """