classifierOF

This function will flag non-worm objects using machine learning.

Usage

classifierOF(data, model = "gbm2x", thresh = 0.6)

Arguments

data

A data frame output from the modelSelection function.

model

Specify one of the built in models. Currently only "gbm2x" is available. This classifier was trained on over 1,000 worm objects selected from a large GWAS experiment and classifies objects as "worm" or "non-worm" with ~90 NOTE: The model was trained to classify poorly segmented worms as 'non-worm' so true worms are often classified as "non-worm".

thresh

The probability threshold for flagging objects based on the classifier. By default the thresh is st to 0.6. Only the objects the classifier predicts to be improperly segmented, with a probability greater than thres, will be flagged.

Value

A single data frame identical to the input data with the classifier_ObjectFlag variable added. The classifier_ObjectFlag variable is coded as "classifier" for objects that are called non-worm by the 2X classifier. All other objects are coded as NA_character, or if there are NAs in any of the variables used to classify objects they are coded as "classErr". The gbm2x_worm_prob variable provides the probability that the object is a properly segmented worm.