Protocol Online logo
Top : New Forum Archives (2009-): : Bioinformatics and Biostatistics

Stats question (exponential growth equation) - (Jan/14/2010 )


Wasnt sure if I should post here or in the cell biology forum. I have a question about exponential growth analysis. I'm analyzing the growth of a cell line in response to a treatment over a 5 day time course. Each treatment group has 6 replicates and I know that the treatment does result in cell death over time. I'm attempting to calculate doubling times (with treatment the doubling time should increase) from the exponential growh equation. My control r2 values are all >.95 but my treatment r2 values are in the .8 to .9 range. My boss tells me that I should analyze my data by maximizing my R2 values (removing any values to result in an increase in my r2). Is that scientifically sound? If I do that I can achieve >.95 R2 on all my treatment groups, however the effect of the treatment becomes negligible. I'm a bit confused :/ Can anyone help?

If the treatment starts to have an effect shouldnt the R2 begin to drop anyways because the relationship is changing?



First; donít be tempted or pressured to remove data points simply to get it looking the way you want. Despite the stories you hear about Mendel itís a practice that will come back to haunt you.
Second; your post sounds like the low R2 is not due to random scatter but rather a growth curve that doesnít follow the exponential equation you are using. Iím not sure what would be the best way of incorporating the toxicity of your compound into the growth equation. Will a logistic function work?

Good luck.