From 0fef3271e6e099d88e9cff81d2d0af2b84946ac4 Mon Sep 17 00:00:00 2001
From: deliaBlue <103108590+deliaBlue@users.noreply.github.com>
Date: Mon, 1 Jan 2024 16:59:15 +0100
Subject: [PATCH] refactor: clean inputs & outputs (#131)

* refactor: remove extra files and file extension

* test: update expected output
---
 test/expected_output.md5   | 2 +-
 workflow/rules/map.smk     | 2 --
 workflow/rules/prepare.smk | 6 +++---
 3 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/test/expected_output.md5 b/test/expected_output.md5
index 5f026bc..6071fd6 100644
--- a/test/expected_output.md5
+++ b/test/expected_output.md5
@@ -47,10 +47,10 @@ be7a0d92e57480190de57eb30baffa36  results/intermediates/extended_mirna_annotatio
 8148cd880602255be166beb59bbed95a  results/intermediates/genome_header.sam
 09e24a504bfec37fee3d5ff1b5c7738e  results/intermediates/exons.bed
 4fb453846e88593d0cac13220ec2d685  results/intermediates/segemehl_genome_index.idx
+44dbf7c3eae00d0bc8d5e1319123746c  results/intermediates/chr_size.tsv
 d34fc868b861b1bc46db07a397dc0f10  results/intermediates/genome_processed.fa.fai
 21e102e4ebd3508bb06f46366a3d578d  results/intermediates/exons.gtf
 003b92b245ac336e3d70a513033e1cee  results/intermediates/transcriptome_trimmed_id.fa
-44dbf7c3eae00d0bc8d5e1319123746c  results/intermediates/chr_size.txt
 cc5c3512dab0e269d82bd625de74198e  results/intermediates/extended_primir_annotation_6_nt.gff3
 f28cc0143ab6659bef3de3a7afa1dccc  results/intermediates/mirna_annotations.gff3
 2d437f8681f4248d4f2075f86debb920  results/intermediates/transcriptome.fa
diff --git a/workflow/rules/map.smk b/workflow/rules/map.smk
index 0c1131f..fe6e34a 100644
--- a/workflow/rules/map.smk
+++ b/workflow/rules/map.smk
@@ -374,7 +374,6 @@ rule map_genome_oligomap:
 rule sort_genome_oligomap:
     input:
         tmap=INTERMEDIATES_DIR / "{sample}" / "oligomap_genome_mappings.fasta",
-        report=INTERMEDIATES_DIR / "{sample}" / "oligomap_genome_report.txt",
         script=SCRIPTS_DIR / "blocksort.sh",
     output:
         sort=INTERMEDIATES_DIR / "{sample}" / "oligomap_genome_sorted.fasta",
@@ -467,7 +466,6 @@ rule map_transcriptome_oligomap:
 rule sort_transcriptome_oligomap:
     input:
         tmap=INTERMEDIATES_DIR / "{sample}" / "oligomap_transcriptome_mappings.fasta",
-        report=INTERMEDIATES_DIR / "{sample}" / "oligomap_transcriptome_report.txt",
         script=SCRIPTS_DIR / "blocksort.sh",
     output:
         sort=INTERMEDIATES_DIR / "{sample}" / "oligomap_transcriptome_sorted.fasta",
diff --git a/workflow/rules/prepare.smk b/workflow/rules/prepare.smk
index 85d5625..db46269 100644
--- a/workflow/rules/prepare.smk
+++ b/workflow/rules/prepare.smk
@@ -53,7 +53,7 @@ rule finish_prepare:
         idx_genome=INTERMEDIATES_DIR / "segemehl_genome_index.idx",
         exons=INTERMEDIATES_DIR / "exons.bed",
         header=INTERMEDIATES_DIR / "genome_header.sam",
-        chrsize=INTERMEDIATES_DIR / "chr_size.txt",
+        chrsize=INTERMEDIATES_DIR / "chr_size.tsv",
         extended_mir=expand(
             INTERMEDIATES_DIR / "extended_mirna_annotation_{extension}_nt.gff3",
             extension=config["extension"],
@@ -320,7 +320,7 @@ rule extract_chr_len:
     input:
         genome=INTERMEDIATES_DIR / "genome_processed.fa.fai",
     output:
-        chrsize=INTERMEDIATES_DIR / "chr_size.txt",
+        chrsize=INTERMEDIATES_DIR / "chr_size.tsv",
     params:
         cluster_log=CLUSTER_LOG / "extract_chr_len.log",
     log:
@@ -339,7 +339,7 @@ rule extract_chr_len:
 rule extend_mirs_annotations:
     input:
         gff3=INTERMEDIATES_DIR / "mirna_annotations.gff3",
-        chrsize=INTERMEDIATES_DIR / "chr_size.txt",
+        chrsize=INTERMEDIATES_DIR / "chr_size.tsv",
         script=SCRIPTS_DIR / "mirna_extension.py",
     output:
         extended_mir=expand(