regression in distToNearest()

[Peter Blazso]

library(shazam)
seqdb <- read.delim("example.tab")
dist <- distToNearest(seqdb, model="ham", first=FALSE, normalize="len" )

The above code worked flawlessly with shazam version 0.1.11, however with 0.2.1 it aborts and throws the following error:

Error in groupGenes(db, v_call = vCallColumn, j_call = jCallColumn, junc_len = NULL, :
one or more of { V_CALL, J_CALL } is factor. Must be character.

If this is intended, please highlight this and the correct way of doing things in the documentation, as well!

Best regards,

Peter

[Julian Zhou]

Thanks for reporting this.

The reason this error gets raised now is because going from v0.2.11 to v0.3.0, additional pre-checks were added to alakazam::groupGenes, which shazam::distToNearest calls, and which checks if the V_CALL and J_CALL columns in the input db are character instead of factor.

With seqdb <- read.delim("example.tab"):

> class(seqdb$V_CALL)
[1] "factor"

Instead, using seqdb <- read.delim("~/Desktop/example.tab", stringsAsFactors=F):

> class(db$V_CALL)
[1] "character"

In short:

• For now, adding stringsAsFactors=F when reading in the input db should solve the problem
• For future, we will handle such conversion internally within shazam::distToNearest, and will add a note of the conversion in the doc. This will appear in the next release.

[Julian Zhou]

Resolved via commit 30868bb

[Peter Blazso]

Now, it's clear. I’ll try your suggestion. Thanks for your help!