src/hg/makeDb/hgLoadWiggle/wigTableStats.sh 1.4
1.4 2009/11/25 19:55:25 hiram
Allow it to work on multiple tables in a single database
Index: src/hg/makeDb/hgLoadWiggle/wigTableStats.sh
===================================================================
RCS file: /projects/compbio/cvsroot/kent/src/hg/makeDb/hgLoadWiggle/wigTableStats.sh,v
retrieving revision 1.3
retrieving revision 1.4
diff -b -B -U 1000000 -r1.3 -r1.4
--- src/hg/makeDb/hgLoadWiggle/wigTableStats.sh 25 Nov 2009 19:10:44 -0000 1.3
+++ src/hg/makeDb/hgLoadWiggle/wigTableStats.sh 25 Nov 2009 19:55:25 -0000 1.4
@@ -1,65 +1,70 @@
#!/bin/sh
# $Id$
DB=$1
T=$2
if [ "x${DB}y" = "xy" -o "x${T}y" = "xy" ]; then
echo "wigTableStats.sh - compute overall statistics for a wiggle table"
echo
- echo "usage: wigTableStats.sh <db> <table>"
- echo "expected table is a wiggle table"
+ echo "usage: wigTableStats.sh <db> <table> [other tables]"
+ echo "expected tables are wiggle tables"
echo "output is a summary of min, max, average, count, sumData, stdDev, viewLimits"
echo "the recommended viewLimits are: mean +- 5*stdDev limited by min,max"
echo "you will want to round those numbers to reasonable nearby values."
exit 255
fi
echo -e "# db.table\tmin max mean count sumData stdDev viewLimits"
+shift # eliminate the database argument
+for T in $*
+do
echo -e -n "${DB}.${T}\t"
hgsql -N ${DB} \
-e "select lowerLimit,dataRange,validCount,sumData,sumSquares from ${T}" \
| awk '
function abs(value) { if (value < 0) {return -value;} else {return value;} }
function viewUpper(min, max, mean, stdDev, fiveDev, range, upper) {
fiveDev = 5 * stdDev;
range = abs(max-min);
upper = mean + fiveDev;
if (upper > max) upper = max;
return upper;
}
function viewLower(min, max, mean, stdDev, fiveDev, range, lower) {
fiveDev = 5 * stdDev;
range = abs(max-min);
lower = mean - fiveDev;
if (lower < min) lower = min;
return lower;
}
BEGIN { lower=3.0e+100; upper=-3.0e+100; count = 0; sumData = 0.0;
sumSquares = 0.0 }
{
maximum = $1 + $2
if ($1 < lower) {lower = $1;}
if (maximum > upper) {upper = maximum;}
count += $3;
sumData += $4;
sumSquares += $5;
}
END {
if (count > 0) {
mean = sumData / count;
var = sumSquares - (sumData*sumData)/count;
stdDev = var;
if (count > 1) { stdDev = sqrt(var/(count-1)); }
vLower = viewLower(lower, upper, mean, stdDev);
vUpper = viewUpper(lower, upper, mean, stdDev);
printf "%g %g %g %d %g %g viewLimits=%g:%g\n",
lower, upper, mean, count, sumData, stdDev, vLower, vUpper
} else {
printf "empty data set\n"
}
}
'
+
+done