| 1 | #!/bin/sh
|
|---|
| 2 |
|
|---|
| 3 |
|
|---|
| 4 | # SETUP
|
|---|
| 5 |
|
|---|
| 6 | # directory where the sequence files are stored
|
|---|
| 7 | sequencedir=/magic/sequences
|
|---|
| 8 | # access to the sql database
|
|---|
| 9 | alias mymysql='mysql -s -u MAGIC -h hercules --password=d99swMT!'
|
|---|
| 10 | # database name
|
|---|
| 11 | db=MyMagic
|
|---|
| 12 |
|
|---|
| 13 |
|
|---|
| 14 | # TODO
|
|---|
| 15 |
|
|---|
| 16 | # Spped up with SQL: FIELD(25123, 1, 2, 3, 4, 5, ..., 1000200);
|
|---|
| 17 | # Cross check sequences in Sequence and other databases
|
|---|
| 18 | # added switch which allows to correct for the problems immediatly
|
|---|
| 19 |
|
|---|
| 20 |
|
|---|
| 21 | # CHECK 1
|
|---|
| 22 | echo Checking if all sequences in Sequences have a corresponding sequence files
|
|---|
| 23 | files=`find $sequencedir -type f`
|
|---|
| 24 | for file in $files
|
|---|
| 25 | do
|
|---|
| 26 | sequence=`echo $file | sed -e "s/^.*\/sequence0\+\([0-9]\+\)\.txt$/\1/"`
|
|---|
| 27 | if [ "$sequence" = "" ]
|
|---|
| 28 | then
|
|---|
| 29 | echo No sequence file: $sequence
|
|---|
| 30 | continue
|
|---|
| 31 | fi
|
|---|
| 32 |
|
|---|
| 33 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql`
|
|---|
| 34 | if ! [ "$sequence" = "$var" ]
|
|---|
| 35 | then
|
|---|
| 36 | echo Sequence-File $sequence exist but it is not in the db.
|
|---|
| 37 | continue
|
|---|
| 38 | fi
|
|---|
| 39 | done
|
|---|
| 40 |
|
|---|
| 41 | # CHECK 2
|
|---|
| 42 | echo Checking if all sequence files have a corresponding entry in Sequences
|
|---|
| 43 | sequences=`echo SELECT fSequenceFirst FROM $db.Sequences | mymysql`
|
|---|
| 44 | for sequence in $sequences
|
|---|
| 45 | do
|
|---|
| 46 | var=`find $sequencedir -regex .*/sequence0+${sequence}\.txt$`
|
|---|
| 47 | if [ "$var" = "" ]
|
|---|
| 48 | then
|
|---|
| 49 | echo Sequence-File for $sequence not found but in db.
|
|---|
| 50 | fi
|
|---|
| 51 | done
|
|---|
| 52 |
|
|---|
| 53 | # CHECK 3
|
|---|
| 54 | echo Checking if all sequences from Sequences exists RunData
|
|---|
| 55 | sequences=`echo SELECT fSequenceFirst FROM $db.Sequences GROUP BY fSequenceFirst | mymysql`
|
|---|
| 56 | for sequence in $sequences
|
|---|
| 57 | do
|
|---|
| 58 | res=`echo SELECT fSequenceFirst FROM $db.RunData WHERE fSequenceFirst=$sequence GROUP BY fSequenceFirst | mymysql`
|
|---|
| 59 | if ! [ "$sequence" = "$res" ]
|
|---|
| 60 | then
|
|---|
| 61 | echo Sequence $sequence exists in Sequences but not in RunData.
|
|---|
| 62 | continue
|
|---|
| 63 | fi
|
|---|
| 64 | done
|
|---|
| 65 |
|
|---|
| 66 | # CHECK 4
|
|---|
| 67 | echo Checking if all sequences from RunData exists in Sequences
|
|---|
| 68 | sequences=`echo SELECT fSequenceFirst FROM $db.RunData WHERE not fSequenceFirst=0 GROUP BY fSequenceFirst | mymysql`
|
|---|
| 69 | for sequence in $sequences
|
|---|
| 70 | do
|
|---|
| 71 | var=`echo SELECT fSequenceFirst FROM $db.Sequences WHERE fSequenceFirst=$sequence | mymysql`
|
|---|
| 72 | if ! [ "$sequence" = "$var" ]
|
|---|
| 73 | then
|
|---|
| 74 | echo Sequence $sequence exists in RunData but not in Sequences.
|
|---|
| 75 | continue
|
|---|
| 76 | fi
|
|---|
| 77 | done
|
|---|
| 78 |
|
|---|