From 25c1cc520dff900c3b34318e003a9774cb90c1f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jo=C3=A3o=20Vitor?= <jvfecav@gmail.com>
Date: Thu, 28 Sep 2023 09:18:28 -0300
Subject: [PATCH] refactor: Gzip DIAMOND outs (#26)

* refactor: Gzip DIAMOND outs

* refactor: Adapt annotate for bytecode text
---
 bin/annotate.py                               | 23 ++++++++++---------
 conf/modules.config                           |  2 +-
 .../diamond/blastx/diamond-blastx.diff        |  9 ++++++++
 modules/nf-core/diamond/blastx/main.nf        |  2 +-
 4 files changed, 23 insertions(+), 13 deletions(-)

diff --git a/bin/annotate.py b/bin/annotate.py
index 9727971..a315ab0 100755
--- a/bin/annotate.py
+++ b/bin/annotate.py
@@ -1,4 +1,5 @@
 #!/usr/bin/env python
+import gzip
 import plyvel
 import os
 import time
@@ -42,7 +43,7 @@ def getAll(
 ):
     check = True
     query = None
-    with open(input, "r") as f:
+    with gzip.open(input, "rb") as f:
         write(out, "Query", "Annotation")
         for line in f:
             ls = line.split(sep)
@@ -73,15 +74,15 @@ def getAll(
             query = ls[queryCol]
             if not checkHit(ls, alen, evalue, bitscore, identity, alenCol, evalueCol, bitscoreCol, pidentCol):
                 if unknown:
-                    write(out, query, "Unknown")
+                    write(out, query.decode(), "Unknown")
                 continue
-            result = db.get(ls[subjectCol].strip().encode())
+            result = db.get(ls[subjectCol].strip())
             if result == None:
                 if unknown:
-                    write(out, query, "Unknown")
+                    write(out, query.decode(), "Unknown")
                 continue
             result = result.decode()
-            write(out, query, result)
+            write(out, query.decode(), result)
 
 
 def getBestHits(
@@ -105,7 +106,7 @@ def getBestHits(
     match = False
     check = True
     query = None
-    with open(input, "r") as f:
+    with gzip.open(input, "rb") as f:
         write(out, "Query", "Annotation")
         for line in f:
             ls = line.split(sep)
@@ -145,19 +146,19 @@ def getBestHits(
             else:
                 if query != ls[queryCol]:
                     if unknown:
-                        write(out, query, "Unknown")
+                        write(out, query.decode(), "Unknown")
                     query = ls[queryCol]
             if not checkHit(ls, alen, evalue, bitscore, identity, alenCol, evalueCol, bitscoreCol, pidentCol):
                 continue
-            result = db.get(ls[subjectCol].strip().encode())
+            result = db.get(ls[subjectCol].strip())
             if result == None:
                 continue
             result = result.decode()
-            write(out, query, result)
+            write(out, query.decode(), result)
             match = True
     if not match:
         if unknown:
-            write(out, query, "Unknown")
+            write(out, query.decode(), "Unknown")
 
 
 def checkHit(ls, alen, evalue, bitscore, identity, alenCol, evalueCol, bitscoreCol, pidentCol):
@@ -269,7 +270,7 @@ def createLevelDB(input, key, value, sep, header, db):
     type=str2bool,
     default=True,
 )
-idmapping_parser.add_argument("--sep", help="The separator between columns (default: \\t)", default="\t")
+idmapping_parser.add_argument("--sep", help="The separator between columns (default: \\t)", default=b"\t")
 
 fixplyvel = subparsers.add_parser("fixplyvel", description="Fix plyvel undefined symbol error by reinstalling it")
 args = parser.parse_args()
diff --git a/conf/modules.config b/conf/modules.config
index e0eb17d..41faf9b 100644
--- a/conf/modules.config
+++ b/conf/modules.config
@@ -78,7 +78,7 @@ process {
 
     // Alignment
     withName: DIAMOND_BLASTX {
-        ext.args = '--more-sensitive --top 3'
+        ext.args = '--more-sensitive --top 3 --compress 1'
         publishDir = [
             path: { "${params.outdir}/alignment/${meta.id}" },
             mode: params.publish_dir_mode,
diff --git a/modules/nf-core/diamond/blastx/diamond-blastx.diff b/modules/nf-core/diamond/blastx/diamond-blastx.diff
index 6135e9e..2655857 100644
--- a/modules/nf-core/diamond/blastx/diamond-blastx.diff
+++ b/modules/nf-core/diamond/blastx/diamond-blastx.diff
@@ -17,5 +17,14 @@ Changes in module 'nf-core/diamond/blastx'
  
      input:
      tuple val(meta), path(fasta)
+@@ -16,7 +18,7 @@
+     output:
+     tuple val(meta), path('*.blast'), optional: true, emit: blast
+     tuple val(meta), path('*.xml')  , optional: true, emit: xml
+-    tuple val(meta), path('*.txt')  , optional: true, emit: txt
++    tuple val(meta), path('*.txt.gz')  , optional: true, emit: txt
+     tuple val(meta), path('*.daa')  , optional: true, emit: daa
+     tuple val(meta), path('*.sam')  , optional: true, emit: sam
+     tuple val(meta), path('*.tsv')  , optional: true, emit: tsv
 
 ************************************************************
diff --git a/modules/nf-core/diamond/blastx/main.nf b/modules/nf-core/diamond/blastx/main.nf
index ccbf8b4..50cf24a 100644
--- a/modules/nf-core/diamond/blastx/main.nf
+++ b/modules/nf-core/diamond/blastx/main.nf
@@ -18,7 +18,7 @@ process DIAMOND_BLASTX {
     output:
     tuple val(meta), path('*.blast'), optional: true, emit: blast
     tuple val(meta), path('*.xml')  , optional: true, emit: xml
-    tuple val(meta), path('*.txt')  , optional: true, emit: txt
+    tuple val(meta), path('*.txt.gz')  , optional: true, emit: txt
     tuple val(meta), path('*.daa')  , optional: true, emit: daa
     tuple val(meta), path('*.sam')  , optional: true, emit: sam
     tuple val(meta), path('*.tsv')  , optional: true, emit: tsv