| 1 | #!/bin/sh | 
|---|
| 2 |  | 
|---|
| 3 |  | 
|---|
| 4 | # SETUP | 
|---|
| 5 |  | 
|---|
| 6 | # directory where the sequence files are stored | 
|---|
| 7 | sequencedir=/magic/sequences | 
|---|
| 8 | callistodir=/magic/data/callisto | 
|---|
| 9 | stardir=/magic/data/star | 
|---|
| 10 | # access to the sql database | 
|---|
| 11 | alias mymysql='mysql -s -u MAGIC -h hercules --password=d99swMT!' | 
|---|
| 12 | # database name | 
|---|
| 13 | db=MyMagic | 
|---|
| 14 |  | 
|---|
| 15 |  | 
|---|
| 16 | # TODO | 
|---|
| 17 |  | 
|---|
| 18 | # Spped up with SQL: FIELD(25123, 1, 2, 3, 4, 5, ..., 1000200); | 
|---|
| 19 | # Cross check sequences in Sequence and other databases | 
|---|
| 20 | # added switch which allows to correct for the problems immediatly | 
|---|
| 21 | # unify the checks (especially 1b and 1c) | 
|---|
| 22 | # for the sequences better use 0* instead of 0+ ? | 
|---|
| 23 |  | 
|---|
| 24 |  | 
|---|
| 25 | # CHECK 1 | 
|---|
| 26 | echo Checking if all sequence files have a corresponding entry in Sequences | 
|---|
| 27 | files=`find $sequencedir -type f` | 
|---|
| 28 | for file in $files | 
|---|
| 29 | do | 
|---|
| 30 | sequence=`echo $file | sed -e "s/^.*\/sequence0\+\([0-9]\+\)\.txt$/\1/"` | 
|---|
| 31 | if [ "$sequence" = "" ] | 
|---|
| 32 | then | 
|---|
| 33 | echo No sequence file: $file | 
|---|
| 34 | continue | 
|---|
| 35 | fi | 
|---|
| 36 |  | 
|---|
| 37 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql` | 
|---|
| 38 | if ! [ "$sequence" = "$var" ] | 
|---|
| 39 | then | 
|---|
| 40 | echo Sequence-File $sequence exist but it is not in Sequences. | 
|---|
| 41 | continue | 
|---|
| 42 | fi | 
|---|
| 43 | done | 
|---|
| 44 |  | 
|---|
| 45 | # CHECK 1b (callisto) | 
|---|
| 46 | echo Checking if all sequences in $callistodir have a corresponding sequence in Sequence | 
|---|
| 47 | dirs=`find $callistodir -type d` | 
|---|
| 48 | for dir in $dirs | 
|---|
| 49 | do | 
|---|
| 50 | sequence=`echo $file | sed -e "s/^.*\/0\+\([0-9]\+\)$/\1/"` | 
|---|
| 51 | if [ "$sequence" = "" ] | 
|---|
| 52 | then | 
|---|
| 53 | echo Invalid directory: $dir | 
|---|
| 54 | continue | 
|---|
| 55 | fi | 
|---|
| 56 |  | 
|---|
| 57 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql` | 
|---|
| 58 | if ! [ "$sequence" = "$var" ] | 
|---|
| 59 | then | 
|---|
| 60 | echo $dir exists but no corresponding sequence in Sequences. | 
|---|
| 61 | continue | 
|---|
| 62 | fi | 
|---|
| 63 | done | 
|---|
| 64 |  | 
|---|
| 65 | # CHECK 1c (star) | 
|---|
| 66 | echo Checking if all sequences in $stardir have a corresponding sequence in Sequence | 
|---|
| 67 | dirs=`find $stardir -type d` | 
|---|
| 68 | for dir in $dirs | 
|---|
| 69 | do | 
|---|
| 70 | sequence=`echo $file | sed -e "s/^.*\/0\+\([0-9]\+\)$/\1/"` | 
|---|
| 71 | if [ "$sequence" = "" ] | 
|---|
| 72 | then | 
|---|
| 73 | echo Invalid directory: $dir | 
|---|
| 74 | continue | 
|---|
| 75 | fi | 
|---|
| 76 |  | 
|---|
| 77 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql` | 
|---|
| 78 | if ! [ "$sequence" = "$var" ] | 
|---|
| 79 | then | 
|---|
| 80 | echo $dir exists but no corresponding sequence in Sequences. | 
|---|
| 81 | continue | 
|---|
| 82 | fi | 
|---|
| 83 | done | 
|---|
| 84 |  | 
|---|
| 85 | # CHECK 2 | 
|---|
| 86 | echo Checking if all sequences in Sequences have a corresponding sequence files | 
|---|
| 87 | sequences=`echo SELECT fSequenceFirst FROM $db.Sequences | mymysql` | 
|---|
| 88 | for sequence in $sequences | 
|---|
| 89 | do | 
|---|
| 90 | var=`find $sequencedir -type f -regex .*/sequence0+${sequence}\.txt$` | 
|---|
| 91 | if [ "$var" = "" ] | 
|---|
| 92 | then | 
|---|
| 93 | echo Sequence-File for $sequence not found but in db. | 
|---|
| 94 | fi | 
|---|
| 95 | done | 
|---|
| 96 |  | 
|---|
| 97 | # CHECK 3 | 
|---|
| 98 | echo Checking if all sequences from Sequences exists RunData | 
|---|
| 99 | sequences=`echo SELECT fSequenceFirst FROM $db.Sequences GROUP BY fSequenceFirst | mymysql` | 
|---|
| 100 | for sequence in $sequences | 
|---|
| 101 | do | 
|---|
| 102 | res=`echo SELECT fSequenceFirst FROM $db.RunData WHERE fSequenceFirst=$sequence GROUP BY fSequenceFirst | mymysql` | 
|---|
| 103 | if ! [ "$sequence" = "$res" ] | 
|---|
| 104 | then | 
|---|
| 105 | echo Sequence $sequence exists in Sequences but not in RunData. | 
|---|
| 106 | continue | 
|---|
| 107 | fi | 
|---|
| 108 | done | 
|---|
| 109 |  | 
|---|
| 110 | # CHECK 4 | 
|---|
| 111 | echo Checking if all sequences from RunData exists in Sequences | 
|---|
| 112 | sequences=`echo SELECT fSequenceFirst FROM $db.RunData WHERE not fSequenceFirst=0 GROUP BY fSequenceFirst | mymysql` | 
|---|
| 113 | for sequence in $sequences | 
|---|
| 114 | do | 
|---|
| 115 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql` | 
|---|
| 116 | if ! [ "$sequence" = "$var" ] | 
|---|
| 117 | then | 
|---|
| 118 | echo Sequence $sequence exists in RunData but not in Sequences. | 
|---|
| 119 | continue | 
|---|
| 120 | fi | 
|---|
| 121 | done | 
|---|
| 122 |  | 
|---|