QPCR analysis with various # of samples from 1 patient? - (Feb/10/2010 )
Dear qPCR friends and experts!
Sorry if I repeat question someone before me posted but I haven't found satisfactory answer to what is my nightmare for several weeks now. I have some previous experiences with relative quantification analysis but what I work on now is a project with tremendously different design and I am completely clueless. 
I search for potential differences in expression of 3 GOIs in a group of patients. The expected differences may appear with the disease progression or patients recovery. That means, I have 1 - 6 specimens from each patient (1 when the patient died soon after being diagnosed, 2 when he seemed to improve (based on clinical parameters), 3 after some time period (that was originally defined when study was designed) etc... 6 specimens is a maximum study wanted to see. Sometimes, however, patients die after 2. sample is draft and sometimes after third....
I run relative quantification, 1 EC, triplicates.
My question is: How can I perform best statistical comparison to get answer to my question “How does GOI 1 (GOI 2, GOI 3) change with disease progression/recovery” when
1) Having various number of samples from each patient (1 – 6)
2) Some patients have only 1 sample (I plan to get samples from healthy individuals, too, so maybe they may serve as some control for these poor patients)
I have no clue what to compare with what, whether to compare all sample from one patient among themselves (but then how to deal with those who died before second blood drawing) and/or to (between/among) other patients…
And as if it were not enough, I always used to do my statistics only with Excel help so I have to admit I have no special software for relative quantification analysis or special stat. software.
Hope I expressed my problem understandably. Thank you for any suggestion.
Paja
Paja on Feb 11 2010, 07:52 AM said:
Sorry if I repeat question someone before me posted but I haven't found satisfactory answer to what is my nightmare for several weeks now. I have some previous experiences with relative quantification analysis but what I work on now is a project with tremendously different design and I am completely clueless.
I search for potential differences in expression of 3 GOIs in a group of patients. The expected differences may appear with the disease progression or patients recovery. That means, I have 1 - 6 specimens from each patient (1 when the patient died soon after being diagnosed, 2 when he seemed to improve (based on clinical parameters), 3 after some time period (that was originally defined when study was designed) etc... 6 specimens is a maximum study wanted to see. Sometimes, however, patients die after 2. sample is draft and sometimes after third....
I run relative quantification, 1 EC, triplicates.
My question is: How can I perform best statistical comparison to get answer to my question “How does GOI 1 (GOI 2, GOI 3) change with disease progression/recovery” when
1) Having various number of samples from each patient (1 – 6)
2) Some patients have only 1 sample (I plan to get samples from healthy individuals, too, so maybe they may serve as some control for these poor patients)
I have no clue what to compare with what, whether to compare all sample from one patient among themselves (but then how to deal with those who died before second blood drawing) and/or to (between/among) other patients…
And as if it were not enough, I always used to do my statistics only with Excel help so I have to admit I have no special software for relative quantification analysis or special stat. software.
Hope I expressed my problem understandably. Thank you for any suggestion.
Paja
Hi there,
Not being sure the type of genes you are examining in your qPCR, this is a tentative suggestion....
Do you have any unaffected control samples that you can run? If you did, you could compare each of your patient sample (& at each time point) back to the control samples, which would be your reference sample. You can then state that you "see a fold-change of X compared to controls, when a patient dies" or that you "see a fold change of X in patients that reach stage 3" etc?
The stats involved with longitudinal data can be very complex (especially when you have data missing) - but you could also examine within each patient the changes that occur as they progress through each stage (1 to 6) if you ran a standard curve with your qPCR, so you can examine your data without a fixed reference sample? You do end up with as many graphs as you have patients using this method though! But might let you see your data and pin point if there's a particular pattern that is occurring?
Data like this is a little hard to interpret, but once you have a 'feel' of what the data is showing, you can pose specific questions and statistically examine the results accordingly.......
Good luck!
