From a52b5cfef1f56a779680a30d04e79fdf5bdf497e Mon Sep 17 00:00:00 2001
From: iquasere <maildosequeira@gmail.com>
Date: Wed, 24 Jan 2024 10:13:36 +0000
Subject: [PATCH] Added minimum_envs folder and contents For commands and
 resources to update envs when needed

---
 resources/minimum_envs/.README             |  8 ++++++++
 resources/minimum_envs/assembly.yaml       | 13 +++++++++++++
 resources/minimum_envs/binning.yaml        |  9 +++++++++
 resources/minimum_envs/de_analysis.yaml    | 15 +++++++++++++++
 resources/minimum_envs/gene_calling.yaml   |  7 +++++++
 resources/minimum_envs/keggcharter.yaml    |  8 ++++++++
 resources/minimum_envs/metaproteomics.yaml |  9 +++++++++
 resources/minimum_envs/normalization.yaml  | 11 +++++++++++
 resources/minimum_envs/preprocess.yaml     | 11 +++++++++++
 resources/minimum_envs/quantification.yaml | 11 +++++++++++
 resources/minimum_envs/recognizer.yaml     |  8 ++++++++
 resources/minimum_envs/reports.yaml        | 12 ++++++++++++
 resources/minimum_envs/seqkit.yaml         |  6 ++++++
 resources/minimum_envs/summary.yaml        |  9 +++++++++
 resources/minimum_envs/upimapi.yaml        |  8 ++++++++
 workflow/scripts/preprocess.py             |  2 --
 16 files changed, 145 insertions(+), 2 deletions(-)
 create mode 100644 resources/minimum_envs/.README
 create mode 100644 resources/minimum_envs/assembly.yaml
 create mode 100644 resources/minimum_envs/binning.yaml
 create mode 100644 resources/minimum_envs/de_analysis.yaml
 create mode 100644 resources/minimum_envs/gene_calling.yaml
 create mode 100644 resources/minimum_envs/keggcharter.yaml
 create mode 100644 resources/minimum_envs/metaproteomics.yaml
 create mode 100644 resources/minimum_envs/normalization.yaml
 create mode 100644 resources/minimum_envs/preprocess.yaml
 create mode 100644 resources/minimum_envs/quantification.yaml
 create mode 100644 resources/minimum_envs/recognizer.yaml
 create mode 100644 resources/minimum_envs/reports.yaml
 create mode 100644 resources/minimum_envs/seqkit.yaml
 create mode 100644 resources/minimum_envs/summary.yaml
 create mode 100644 resources/minimum_envs/upimapi.yaml

diff --git a/resources/minimum_envs/.README b/resources/minimum_envs/.README
new file mode 100644
index 0000000..6e8b4ff
--- /dev/null
+++ b/resources/minimum_envs/.README
@@ -0,0 +1,8 @@
+This folder contains the minimum tools required for each MOSCA environment.
+
+When updating a new environment, the following commands should be run from this directory, assigning the correct value for the `ENV_NAME` variable:
+`bash
+ENV_NAME=env_name
+mamba create $ENV_NAME.yml
+mamba env export --from-history -f $ENV_NAME.yml > $ENV_NAME.yml
+`
\ No newline at end of file
diff --git a/resources/minimum_envs/assembly.yaml b/resources/minimum_envs/assembly.yaml
new file mode 100644
index 0000000..eecfc9f
--- /dev/null
+++ b/resources/minimum_envs/assembly.yaml
@@ -0,0 +1,13 @@
+name: assembly
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - bowtie2
+  - gmcloser
+  - megahit
+  - quast
+  - samtools
+  - spades
diff --git a/resources/minimum_envs/binning.yaml b/resources/minimum_envs/binning.yaml
new file mode 100644
index 0000000..c7e787f
--- /dev/null
+++ b/resources/minimum_envs/binning.yaml
@@ -0,0 +1,9 @@
+name: binning
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - checkm-genome
+  - maxbin2
+  - pandas
diff --git a/resources/minimum_envs/de_analysis.yaml b/resources/minimum_envs/de_analysis.yaml
new file mode 100644
index 0000000..e824f20
--- /dev/null
+++ b/resources/minimum_envs/de_analysis.yaml
@@ -0,0 +1,15 @@
+name: de_analysis
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - bioconductor-annotationdbi      # check if needed
+  - bioconductor-deseq2
+  - bioconductor-genomeinfodb
+  - bioconductor-genomeinfodbdata   # check if needed
+  - bioconductor-genomicranges      # check if needed
+  - bioconductor-rots
+  - r-base=4                        # check if needed
+  - r-pheatmap
diff --git a/resources/minimum_envs/gene_calling.yaml b/resources/minimum_envs/gene_calling.yaml
new file mode 100644
index 0000000..5108d12
--- /dev/null
+++ b/resources/minimum_envs/gene_calling.yaml
@@ -0,0 +1,7 @@
+name: gene_calling
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - fraggenescan
diff --git a/resources/minimum_envs/keggcharter.yaml b/resources/minimum_envs/keggcharter.yaml
new file mode 100644
index 0000000..cd02783
--- /dev/null
+++ b/resources/minimum_envs/keggcharter.yaml
@@ -0,0 +1,8 @@
+name: keggcharter
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - keggcharter>=1.1.2
diff --git a/resources/minimum_envs/metaproteomics.yaml b/resources/minimum_envs/metaproteomics.yaml
new file mode 100644
index 0000000..e089a32
--- /dev/null
+++ b/resources/minimum_envs/metaproteomics.yaml
@@ -0,0 +1,9 @@
+name: metaproteomics
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - peptide-shaker=2
+  - searchgui=4
diff --git a/resources/minimum_envs/normalization.yaml b/resources/minimum_envs/normalization.yaml
new file mode 100644
index 0000000..5e5fb33
--- /dev/null
+++ b/resources/minimum_envs/normalization.yaml
@@ -0,0 +1,11 @@
+name: normalization
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - bioconductor-limma          # check if needed
+  - bioconductor-pcamethods
+  - bioconductor-vsn
+  - r-base=4                    # check if needed
diff --git a/resources/minimum_envs/preprocess.yaml b/resources/minimum_envs/preprocess.yaml
new file mode 100644
index 0000000..f9ef25f
--- /dev/null
+++ b/resources/minimum_envs/preprocess.yaml
@@ -0,0 +1,11 @@
+name: preprocess
+channels:
+  - bioconda
+  - conda-forge
+  - defaults
+dependencies:
+  - fastqc
+  - pandas
+  - numpy
+  - sortmerna=4
+  - trimmomatic=0.39
diff --git a/resources/minimum_envs/quantification.yaml b/resources/minimum_envs/quantification.yaml
new file mode 100644
index 0000000..e0a9754
--- /dev/null
+++ b/resources/minimum_envs/quantification.yaml
@@ -0,0 +1,11 @@
+name: quantification
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - bowtie2
+  - numpy
+  - pandas
+  - samtools
diff --git a/resources/minimum_envs/recognizer.yaml b/resources/minimum_envs/recognizer.yaml
new file mode 100644
index 0000000..03dbe9e
--- /dev/null
+++ b/resources/minimum_envs/recognizer.yaml
@@ -0,0 +1,8 @@
+name: recognizer
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - recognizer>=1.10.1
diff --git a/resources/minimum_envs/reports.yaml b/resources/minimum_envs/reports.yaml
new file mode 100644
index 0000000..23d3e22
--- /dev/null
+++ b/resources/minimum_envs/reports.yaml
@@ -0,0 +1,12 @@
+name: reports
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - numpy
+  - openpyxl
+  - pandas
+  - tqdm
+  - xlsxwriter
diff --git a/resources/minimum_envs/seqkit.yaml b/resources/minimum_envs/seqkit.yaml
new file mode 100644
index 0000000..9b964de
--- /dev/null
+++ b/resources/minimum_envs/seqkit.yaml
@@ -0,0 +1,6 @@
+name: seqkit
+channels:
+  - bioconda
+  - defaults
+dependencies:
+  - seqkit
diff --git a/resources/minimum_envs/summary.yaml b/resources/minimum_envs/summary.yaml
new file mode 100644
index 0000000..dc834fb
--- /dev/null
+++ b/resources/minimum_envs/summary.yaml
@@ -0,0 +1,9 @@
+name: summary
+channels:
+  - conda-forge
+  - defaults
+dependencies:
+  - numpy
+  - openpyxl
+  - pandas
+  - tqdm
diff --git a/resources/minimum_envs/upimapi.yaml b/resources/minimum_envs/upimapi.yaml
new file mode 100644
index 0000000..8ddcb90
--- /dev/null
+++ b/resources/minimum_envs/upimapi.yaml
@@ -0,0 +1,8 @@
+name: upimapi
+channels:
+  - conda-forge
+  - bioconda
+  - anaconda
+  - defaults
+dependencies:
+  - upimapi>=1.13.1
diff --git a/workflow/scripts/preprocess.py b/workflow/scripts/preprocess.py
index 883231e..a924558 100644
--- a/workflow/scripts/preprocess.py
+++ b/workflow/scripts/preprocess.py
@@ -9,10 +9,8 @@
 March 2017
 '''
 
-import argparse
 import shutil
 from glob import glob
-from multiprocessing import cpu_count
 import os
 import pathlib
 import time