Swap plyr for dplyr

[Jason Vander Heiden]

Both alakazam and tigger have been moved to dplyr. This will surely cause problems unless we either ensure all plyr functions are called using package references (plyr::) or we replace the plyr calls with corresponding dplyr calls.

Note the message upon load of shm:

You have loaded plyr after dplyr - this is likely to cause problems.
If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
library(plyr); library(dplyr)

We should also swap reshape2 for tidyr.

Alternatively, we could swap the plyr/reshape2 functions for data.table equivalents. I have no preference.

[ssnn]

#ed4d5dd

[Jason Vander Heiden]

I think we need to go some type casting to remove warnings. I'm getting these from calcBaseline() and groupBaseline():

Warning messages:
1: In rbind_all(x, .id) : Unequal factor levels: coercing to character

Also, from the bind_rows man page:

rbind_list and rbind_all have been deprecated. Instead use bind_rows

[ssnn]

I don't get this warning with the example code for calcBaseline. Are you using dplyr_0.4.3.?

library("alakazam")
library("shm")
dbPath <- system.file("extdata", "Influenza.tab", package="shm")
db <- readChangeoDb(dbPath)
db_baseline <- calcBaseline(db,
                            sequenceColumn="SEQUENCE_IMGT",
                            germlineColumn="GERMLINE_IMGT_D_MASK",
                            testStatistic="focused",
                            regionDefinition=IMGT_V_NO_CDR3,
                            targetingModel = HS5FModel,
                            nproc = 6)

But I do get the warning with groupBaseline, and the source seems to be summarizeBaseline(baseline):

dplyr::bind_rows(df_baseline_seq, df_baseline_seq_region)

when binding CDR and FWR regions. I have to investigate.

Is this a bug in dplyr? bind_rows calls rbind_all, which is deprecated:

bind_rows
function (..., .id = NULL) 
{
    dots <- list(...)
    if (is.list(dots[[1]]) && !is.data.frame(dots[[1]]) && !length(dots[-1])) {
        x <- dots[[1]]
    }
    else {
        x <- dots
    }
    if (!is.null(.id)) {
        if (!(is.character(.id) && length(.id) == 1)) {
            stop(".id is not a string", call. = FALSE)
        }
        names(x) <- names(x) %||% seq_along(x)
    }
    rbind_all(x, .id)
}
<environment: namespace:dplyr>

[ssnn]

dfb3540 fixes the groupBaseline() warning.

[ssnn]

abde716