<html><div style='background-color:'><DIV><FONT face=Courier>Hi all,<BR><BR>I have just started the microarray analysis of 4 twin pairs. Each twin pair consists of 1 normal and 1 affected (high levels of Fetal Haemaglobin). I am having some problems deciding how to normalise this data. </FONT></DIV>
<DIV><FONT face=Courier></FONT> </DIV>
<DIV><FONT face=Courier>Should I normalize each pair of twins together (as well as normalizing separately) and analyse each pair of twins for differences, with the hope of finding correlations across the group. Or, would it be better to analyse the data as a whole in two groups (affected/unaffected). Wouldn't this increase the variation and therefore make it more difficult to find the gene(s) causing this trait?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=Courier>We are using erythroid cultures that have been sorted by MACS to 95%+ purity. We are using Affymetrix chips and have the demo version of GeneSpring (with no hope of purchasing it). We have also downloaded Bioconductor and R.</FONT></DIV>
<P><FONT face=Courier>We have had some problems with normalizing in GeneSpring. Has anyone else had problems normalizing two samples? The options are limited and normalizing to a median give a symmetrical result and normalizing to one sample makes the expression values for that sample all 1 (obviously). It seems ridiculous that you cannot load data that has been prenormalized (or have we missed something)?<BR><BR>I would really appreciate any help/advice that people can give as we have little support in our lab.</FONT></P>
<P><FONT face=Courier>Thanks,</FONT></P>
<P><FONT face=Courier>Catherine</FONT></P>
<DIV><BR><BR></DIV></div><br clear=all><hr>Protect your PC from e-mail viruses. <a href="http://g.msn.com/8HMHENUK/2740??PS=">Get MSN 8 today.</a> </html>