Statistical help required urgently

[piemann]

Hi all,

I have two absolute quantitation assays that I use to generate a readout of copies of target per pg of input RNA. Both assays generate a standard deviation (each sample is done in triplicate), one in copies per reaction (target assay) and one in pg of RNA per reaction (endogenous assay).

Is there a method of combining the two standard deviations to get an overall error value for the assay? They are different units, so I am uncertain if this is even possible. If it cannot be done this way, is there a statistical conversion I can do to demonstrate any kind of standard error/deviation?

Thanks,

Iain.

[danfive]

I think you want CV.
see ---> [Only registered users see links. ]
I does not combine your two std deviations, instead you will pick one and turn it into percent of the mean.
This makes it more readable to say 55ng + or - 10% instead of 55ng + or - 5.5ng.

[piemann]

Hi danfive, thanks for your reply. I am still a bit unsure of this though.

Can i use the mean square error of the two cvs to get an overall standard error for the assay.

As an example,

Sample 1 contains 1e6+/-2e4 copies per reaction of my viral target
Sample 2 contains 1e5+/-3e3 pg of total RNA (GAPDH endogenous QPCR)

I would represent this as 10 copies of target per pg of RNA

Using CV:
The CV for the target assay is (2e4/1e6)*100 = 2%
The CV for the endogenous assay is (3e3/1e5)*100 = 3%

Can I calculate a mean square error using the formula?

mse = sqrt((cv1*cv1)+(cv2*cv2))
= sqrt(4+9)
= 3.0655%

Therefore my result would be 10 copies per pg of total RNA +/- 3.07%


Is this a valid use of mean square error?

Any and all help gratefully received.


Thanks,
Iain.

[danfive]

If I have this right:
Assay 1: Tells you absolute copy number of Target gene
Assay 2: Tells you pg RNA of GAPDH gene

If so you want to use target gene to reference gene normalization.

This may help, read sections for absolute quantification and normalization of expression
---> [Only registered users see links. ]

You may also want to look into this reference on the issue of error bars in experimental biology ---> [Only registered users see links. ]