behavr (version 0.3.2)

behavr: An S3 class, based on data.table, to store ethomics data

Description

In modern behavioural biology, it is common to record long time series of several variables (such as position, angle, fluorescence and many others) on multiple individuals. In addition to large multivariate time series, each individual is associated with a set of metavariables (i.e. sex, genotype, treatment and lifespan ), which, together, form the metadata. Metavariables are crucial in so far as they generally "contain" the biological question. During analysis, it is therefore important to be able to access, alter and compute interactions between both variables and metavariables. behavr is a class that facilitates manipulation and storage of metadata and data in the same object. It is designed to be both memory-efficient and user-friendly. For instance, it abstracts joins between data and metavariables.

Usage

behavr(x, metadata)
setbehavr(x, metadata)
is.behavr(x)

Arguments

x

data.table containing all measurements

metadata

data.table containing the metadata

Details

A behavr table is a data.table. Therefore, it can be used by any function that would work on a data.frame or a data.table. Most of the operation such as variable creation, subsetting and joins are inherited from the data.table [] operator, following the convention DT[i,j,by] (see data table package for detail). These operations are applied on the data. Metadata can be accessed using meta=TRUE: DT[i,j,by, meta=TRUE], which allows extraction of subsets, creation of metavariables, etc.

Both x and metadata should have a column set as key with the same name (typically named id). behavr() copies x, whilst setbehavr() uses reference. metadata is always copied.

References

• The relevant rethomic tutorial section -- about metavariables and variables in this context

See Also

• data.table -- on which behavr is based

• xmv -- to join metavariables

• rejoin -- to join all metadata

• bind_behavr_list -- to merge several behavr tables

Examples

# NOT RUN {
# We generate some metadata and data
set.seed(1)
met <- data.table::data.table(id = 1:5,
                              condition = letters[1:5],
                              sex = c("M", "M", "M", "F", "F"),
                              key = "id")
data <- met[  ,
              list(t = 1L:100L,
                  x = rnorm(100),
                  y = rnorm(100),
                  eating = runif(100) > .5 ),
              by = "id"]
# we store them together in a behavr object d
# d is a copy of the data
d <- behavr(data, met)
print(d)
summary(d)

# we can also convert data to a behavr table without copy:
setbehavr(data, met)
print(data)
summary(data)

### Operations are just like in data.table
# row subsetting:
d[t < 10]
# column subsetting:
d[, .(id, t, x)]
# making new columns inline:
d[, x2 := 1 - x]
### Using `meta = TRUE` applies the operation on the metadata
# making new metavariables:
d[, treatment := interaction(condition,sex), meta = TRUE]
d[meta = TRUE]

# }