# 2D image analysis using Image Master software - How to interpret data??? (Apr/26/2006 )

Hi all...

I am carrying out 2D image analysis using Image Master(Amersham Bioseciences) software.Instead of analysing whole gels we are analysing part of gels,say containing 10 spots.I need to know how to interpret and present the data.I have values for Volume,%Volume,Intensity, %Intensity.which parameter out of these four is used to interpret the data and why?i need to know it for analysing my data....also i need to know what is meant by normalisation of spots?can somebody suggest me please....

thanks.......

-transposon9-

QUOTE (transposon9 @ Apr 26 2006, 08:29 AM)
Hi all...

I am carrying out 2D image analysis using Image Master(Amersham Bioseciences) software.Instead of analysing whole gels we are analysing part of gels,say containing 10 spots.I need to know how to interpret and present the data.I have values for Volume,%Volume,Intensity, %Intensity.which parameter out of these four is used to interpret the data and why?i need to know it for analysing my data....also i need to know what is meant by normalisation of spots?can somebody suggest me please....

thanks.......

Normalisation is maths! (euugh)
by normalising you make you values sum to one (or another value if so so desire (?standardisation?)) but retain the same ratios to one another.

This code will work if put in a file and run using perl.
CODE
#Supply a list of values in an array.
@foo = (0, 1, 2, 3, 44, 97, 12);

@norm =norm(\@foo);
for (\$i = 0; \$i < @norm; \$i++) {
print "\$foo[\$i] :: \$norm[\$i]\n"; #THIS WILL PRINT BEFORE :: AFTER VALUE
}

sub norm {
my (\$ref) = @_;
my (\$tot, @result);
for(@\$ref) {\$total += \$_ }
for(@\$ref) {push (@result, \$_ / \$total) }
return (@result);
}

Most of the papers I have read (some time ago on 'real' (as opposed to theoretical) list the relative intensity - for obvious reasons (it allows for direct comparison in a quantitative fashion

-perlmunky-

I don't think this code will work as written...

@norm contains a single item -- a reference to the @foo array. When the print statement executes (which it will do only once) it will return the value of \$foo[\$i] with \$i = 0, and an array reference (since that's what \$norm[0] is). So, you'll get something like:

0 :: ARRAY(0x225dc0)

and that's it...

Plus, norm() is never called, so the normalization routine is never executed.

transposon9 -- Normalization can be performed in many ways, depending on the data and what it is you're trying to do. One can use quantile normalization, log2 ratios, simple ratios, etc.

There is also some confusion about what "normalization" means. Originally, normalization meant to transform groups of data so they were directly comparable to one another. Say one laboratory measured people's height in inches, and another measured (other) people's height in centimeters. To make these data comparable, one would need to multiply all the data in the inches group by 2.54 (or multiply the centimeter group's data by 0.393700787) before the groups could be prepared. This, to me anyhow, is "standardization".

Other definitions of normalization are used to minimize systematic bias in data (variation due to non-biological effects). This type of normalization comes up quite a bit in microarray studies. Much of the normalization in microarray studies is also aimed at allowing one set of data to be compared to another (microarrays are the biostatistician's full employment act ) and removing systemic bias, but is concerned with making the data "normal". Many of the common statistical techniques such as ANOVA, t-, z- and F-tests, etc. assume your data is normally distributed. Normalization in this case is treating the data so that it falls within a normal distribution, and thus the assumption of these significance tests can be validated.

BTW, don't get me wrong -- no one would ever confuse me with a statistician...

I assume your need to to normalize spot intensity data so you can compare data on one 2D gel (say control conditions) to that gathered from another gel (say, experimental conditions)?

-HomeBrew-

That's the second time this week I have been calling functions withouth the function name - too much coding this week - switching between perl, c++, java and now ruby too! Arrgh my head hurts.

Every time I mention what I think is standardisation I get told there is no such thing and it's all normalisation! Conversations usually go:

ME> Ok I have this data and I want to compare it to X so I will standardise it.
POST-DOC> huh, you mean normalisation
ME> really? OK
POST-DOC> what is standardisation?

To make life that bit more interesting see:
http://en.wikipedia.org/wiki/

Oh just found this:
http://www.ivorix.com/en/products/tech/norm/index.html
Looks good at first scan.

-sorry for the perl mistake.

-perlmunky-