annotate 0.5.0/dbcheck @ 0:3c767f9cfd88 draft default tip

planemo upload
author galaxytrakr
date Fri, 29 May 2026 13:37:56 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
1 #!/usr/bin/env bash
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
2
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
3 ##########################################################
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
4 # Constants
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
5 ##########################################################
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
6 GREEN=$(tput setaf 2)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
7 RED=$(tput setaf 1)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
8 CYAN=$(tput setaf 6)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
9 CLRESET=$(tput sgr0)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
10 prog_name="nowayout"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
11 dbBuild="03062025"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
12 SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
13 dbPath="${SCRIPT_DIR}/assets/dbfiles"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
14 taxonomyPath="$dbPath/taxonomy"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
15
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
16 usage()
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
17 {
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
18 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
19 echo usage: "$0" [-h]
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
20 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
21 echo "Check for species presence in ${prog_name} database(s)."
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
22 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
23 echo 'Example usage:'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
24 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
25 echo 'dbcheck -l'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
26 echo 'dbcheck -g Cathartus'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
27 echo 'dbcheck -d mitomine -g Cathartus'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
28 echo 'dbcheck -d mitomine -s "Cathartus quadriculus"'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
29 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
30 echo 'Options:'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
31 echo " -l : List ${prog_name} databases"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
32 echo ' -d : Search this database. Default: mitomine.'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
33 echo ' -g : Genus to search for.'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
34 echo ' -s : "Genus Species" to search for.'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
35 echo ' -h : Show this help message and exit'
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
36 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
37 echo "$1"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
38 }
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
39
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
40 while getopts ":d:g:s:l" OPT; do
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
41 case "${OPT}" in
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
42 l)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
43 listdb="list"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
44 ;;
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
45 d)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
46 dbname=${OPTARG}
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
47 ;;
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
48 g)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
49 genus=${OPTARG}
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
50 ;;
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
51 s)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
52 species=${OPTARG}
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
53 ;;
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
54 ?)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
55 usage
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
56 exit 0
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
57 ;;
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
58 esac
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
59 done
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
60
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
61
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
62
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
63 if [ -n "$listdb" ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
64 num_dbs=$(find -L "$taxonomyPath" -type d | tail -n+2 | wc -l)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
65 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
66
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
67 db_num="1"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
68 find -L "$taxonomyPath" -type d | tail -n+2 | while read -r db; do
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
69 dbName=$(basename "$db")
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
70 echo "${db_num}. $dbName"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
71 db_num=$(( db_num + 1 ))
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
72 done
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
73 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
74 echo "Number of ${prog_name} databases: $num_dbs"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
75 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
76
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
77 exit 0
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
78 fi
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
79
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
80
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
81
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
82 if [ -z "$dbname" ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
83 dbname="mitomine2"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
84 fi
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
85
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
86 if [[ -n "$genus" && -n "$species" ]]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
87 usage "ERROR: Only one of -g or -s needs to be defined!"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
88 exit 1
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
89 elif [ -n "$genus" ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
90 check="$genus"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
91 elif [ -n "$species" ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
92 check="$species"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
93 else
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
94 check=""
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
95 fi
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
96
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
97 if [ -z "$check" ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
98 usage "ERROR: -g or -s is required! check:$check"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
99 exit 1
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
100 fi
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
101
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
102 lineages="$taxonomyPath/$dbname/lineages.csv"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
103
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
104 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
105 echo -e "Checking ${dbname} for ${CYAN}${check}${CLRESET}...\nPlease wait..."
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
106 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
107
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
108 num=$(grep -F ",$check," "$lineages" | cut -f1 -d, | sort -u | wc -l)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
109 num_species=$(tail -n+2 "$lineages" | cut -f8 -d, | sort -u | wc -l)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
110 num_entries=$(tail -n+2 "$lineages" | wc -l)
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
111
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
112 echo "$dbname brief stats"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
113 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
114 echo "DB Build: $dbBuild"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
115 echo "Number of unique species: $num_species"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
116 echo "Number of accessions in database: $num_entries"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
117 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
118
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
119
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
120 if [ "$num" -gt 0 ]; then
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
121 echo
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
122 echo "${GREEN}$check is present in ${dbname}${CLRESET}."
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
123 echo "Number of accessions representing $check: $num"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
124 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
125 else
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
126 echo "${RED}$check is absent in ${dbname}${CLRESET}."
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
127 echo -e "No worries. Please request the developer of\n${prog_name} to augment the database!"
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
128 echo "=============================================="
3c767f9cfd88 planemo upload
galaxytrakr
parents:
diff changeset
129 fi