unifiedWMWqPCR (version 1.8.0)

getEstimate: Extract logor, or, odds and percentages from a uwmwRes object

Description

With this function you can extract the modelled (log) odds ratio, odds or percentages that represent the chance on differential expression as estimated by the uWMW function. It also allows to extract either the standard error of or the confidence interval around the estimates. See Details for more explanation.

Usage

"getEstimate"(x, esttype = c("logor", "or", "logodds", "odds", "p"), se.fit = (esttype %in% c("logor", "logodds")), ci = TRUE, drop = TRUE)

Arguments

x

an object of the clas uwmwRes

esttype

a character string indicating the measure you want to extract. It can take the values logor for the log odds ratio, or for the odds ratios, logodds for the log odds, odds for the odds or p for the percentages.

se.fit

logical value indicating whether the standard errors of the logor or the log odds should be returned as well. Ignored when type has a value different from logor or logodds. Note that you can also use the accessor se to get only the standard errors.

ci

numerical value indicating the confidence interval (0.95 is 95% confidence interval). If set to TRUE, the 95 NULL, no confidence interval is returned.

drop

a logical value. If set to TRUE and neither se nor ci is calculated, the function returns a vector instead of an uwmwEstimate object.

...

passes on arguments to the next method

Value

In general, a uwmwEstimate object with the requested estimate. See uwmwEstimate for details. In case drop=TRUE and neither the standard error nor the confidence interval is calculated, a numeric named vector.

Details

The function can only calculate standard errors for the log OR and the log odds. In all other cases, se.fit is ignored. The function takes into account a possible ordering in the object (see also sort.uwmwRes). So take into account that you get the estimates in the specified order. In case you want this different, either use the function unorder on the object first, or check if any of the uwmw_Accessors can help you out.

The argument se.fit is mainly to be used to save calculation time. Normally there's no need to set it to FALSE.

Examples

data(NBmat)
NBtest <- uWMW(NBmat, groups=NBgroups)
getEstimate(NBtest,'logodds')
getEstimate(NBtest,'odds',ci=0.9)