WIP: Checkpoint snakefile

2019_GDB/config/CONFIG.yaml

-# steps: "assembly_annotation mapping metaT mmseq binning taxonomy analysis"
-steps: "binning taxonomy"
+steps: ["assembly_annotation", "mapping", "metaT", "mmseq", "binning", "taxonomy", "analysis"]
+
+# analysis_steps: ["cdhit", "mappability", "crispr", "plasmids", "amr"]
+analysis_steps: ["cdhit", "mappability", "crispr", "plasmids", "amr"]
+
+# working directory containing all relevant data,
+# i.e. prefix for data, results, DBs etc.
+work_dir: "/scratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB"
 data_dir: "data"
 results_dir: "results"
 db_dir: "dbs"
+
 runs:
     first: "S1_SizeSelected"
     second: "S3_Gtube"
 #    third: "20181108_0827_test"
-assemblers: ["flye"]
+
+# assemblers: ["flye"]
+assemblers: ["flye", "megahit", "metaspades", "metaspades_hybrid"]
+
 p7zip:
     bin: "/home/users/claczny/apps/software/p7zip_16.02/bin/7za"
     threads: 4
@@ -18,8 +28,10 @@ ont_fast5_api:
         threads: 8
 flowcell: "FLO-MIN106"
 kit: "SQK-LSK109"
+
 #barcodes: ["barcode06", "barcode07", "barcode08", "barcode09", "barcode10"]
 barcodes: ["no_barcode"]
+
 guppy_cpu:
     path: "/scratch/users/claczny/ont/apps/software/ont-guppy-cpu-3.1.5_linux64/bin"
     bin: "/scratch/users/claczny/ont/apps/software/ont-guppy-cpu-3.1.5_linux64/bin/guppy_basecaller"
@@ -45,10 +57,13 @@ guppy_barcoder:
     version: "3.4.5+fb1fbfb"
     records_per_fastq: 8000
     threads: 8
+
 nanostats:
+
 short_reads_prefix: "/scratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB/data/raw/short_reads"
 #short_reads_prefix: "/mnt/isilon/projects/lcsb_sequencing/transfer/bioecosystem/Rashi/2019/Apr/fastq"
 metaT_prefix: "/scratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB/data/metaT"
+
 fastp:
     min_length: 40
 minimap2:
@@ -91,15 +106,15 @@ rebaler:
     threads: 28
 diamond:
     threads: 28
-    db: "/mnt/isilon/projects/ecosystem_biology/NOMIS/DIAMOND/new_nr.dmnd"
+    #db: "/mnt/isilon/projects/ecosystem_biology/NOMIS/DIAMOND/new_nr.dmnd"
+    db: "/work/projects/ecosystem_biology/local_tools/databases/nr_uniprot_trembl.dmnd"
 metaspades:
     threads: 28
 mmseq2:
     threads: 24
+
 # Define sample names
-#samples: ["flye", "megahit", "metaspades_hybrid"]
-# samples: ["flye", "megahit"]
-# samples: ["metaspades_hybrid"]
+samples: ["ONT3_MG_xx_Rashi_S11"]
 binning_samples: ["flye", "megahit", "bwa_sr_metaspades_hybrid", "bwa_lr_metaspades_hybrid", "bwa_merged_metaspades_hybrid", "mmi_sr_metaspades_hybrid", "mmi_lr_metaspades_hybrid", "mmi_merged_metaspades_hybrid"]
 
 # Hybrid assembler
@@ -119,10 +134,11 @@ kraken2_database: "/scratch/users/bkunath/Kraken2/maxikraken2_1903_140GB/"
 
 # Path to DAS_Tool
 DAS_Tool:
-        path: "/home/users/sbusi/apps/DAS_Tool-master"
-        bin: "/home/users/sbusi/apps/DAS_Tool-master/src/"
+    path: "/home/users/sbusi/apps/DAS_Tool-master"
+    bin: "/home/users/sbusi/apps/DAS_Tool-master/src/"
 
 # Path to DAS_Tool database
+# TODO: mv to DAS_Tool
 dastool_database: "/home/users/sbusi/apps/DAS_Tool-master/db/"
 
 # Mapping options
@@ -144,10 +160,29 @@ GTDBTK:
     DATA: "/home/users/sbusi/apps/db/gtdbtk/release89"
 
 # Rscript path
+# TODO: mv to DAS_Tool
 Rscript: "/home/users/sbusi/apps/miniconda3/envs/dastool/bin/"
 
+# XXX
 mmseqs:
-     path: "/home/users/sbusi/apps/mmseqs/bin"
-     createdb: "/home/users/sbusi/apps/mmseqs/bin/mmseqs createdb"
-     rbh: "/home/users/sbusi/apps/mmseqs/bin/mmseqs rbh"
-     convertalis: "/home/users/sbusi/apps/mmseqs/bin/mmseqs convertalis"
+    path: "/home/users/sbusi/apps/mmseqs/bin"
+    createdb: "/home/users/sbusi/apps/mmseqs/bin/mmseqs createdb"
+    rbh: "/home/users/sbusi/apps/mmseqs/bin/mmseqs rbh"
+    convertalis: "/home/users/sbusi/apps/mmseqs/bin/mmseqs convertalis"
+
+# CRISPR
+CASC:
+    PATH: "$PATH:/mnt/lscratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB/crispr/bin"
+    PERL5LIB: "/mnt/lscratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB/crispr/lib/site_perl"
+minced:
+    PATH: "$PATH:/mnt/lscratch/users/sbusi/ONT/cedric_ont_basecalling/2019_GDB/crispr/minced/"
+
+# Plasmid prediction
+plasflow:
+    threshold: 0.7 # class. prob. threshold
+    minlen: 1000 # rm contigs with length below this threshold
+
+# AMR prediction
+rgi:
+    db_url: "https://card.mcmaster.ca/latest/data"
+    alignment_tool: "DIAMOND" # DIAMOND or BLAST