I'm working with a dataset consisting of degradation rates for proteins in an organism, a total of approx 3750 rates in total. Obtaining these rates is difficult, so the majority of rates are reported with only n=1, however a small subset have been reported with n=2, so the sample standard deviation for each protein can be (poorly) estimated from this.

The rates are sampled from a large population of cells so, it's reasonable to say that any variance is experimental rather than due to variance between cells.

My question is: if all the errors are due to experimental error, and thus probably drawn from the same distribution, is it reasonable to attempt to infer the error distribution and then extrapolate this to the proteins whose rate has only been sampled once?

Alternatively, am I missing a trick, and should really be doing something completely different! I don't have much formal statistical training, so am at a bit of a loss here.

Thanks in advance.


It is not possible to calculate a standard deviation based on an N of 2. If some of your proteins are very similar, perhaps you could combine the data you have for them and then use the resulting larger data sets (with N of 3 or more) to estimate the error variance. Perhaps that would give you an overestimate, but some estimate might be better than no estimate. Alternatively, you might consider gathering a 3rd data point for some of the proteins for which you already have 2 data points.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.