Add PlasFlow and RGI

merge checkpoint_snakefile_VG into checkpoint_snakefile to include PlasFlow and RGI

Note: The config file and main snakemake file have been adjusted. Further changes should be made in other snakemake files as only the files for the analysis step have been adjusted.

See commits mentioned in the issue #28 (closed) and #29 (closed).

2019_GDB/config/CONFIG.yaml

 # steps: "assembly_annotation mapping metaT mmseq binning taxonomy analysis"
-steps: "binning taxonomy analysis"
+# steps: "binning taxonomy analysis"
+steps: "analysis"
+# analysis_steps: ["cdhit", "mappability", "crispr", "plasmids", "amr"]
+analysis_steps: ["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"
@@ -7,7 +13,8 @@ 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
@@ -153,10 +160,19 @@ mmseqs:
      rbh: "/home/users/sbusi/apps/mmseqs/bin/mmseqs rbh"
      convertalis: "/home/users/sbusi/apps/mmseqs/bin/mmseqs convertalis"
 
-# Paths to CASC and minced for CRISPR
+# 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