Spencer Nystrom

This same thing has bit me a couple times. My default assumption about how a tool like this should behave is that it would do no filtering unless asked. q60...

This works and is probably the easiest solution: ``` pull.GenomicRanges(x, ...){ dplyr::pull(data.frame(x), ...) } ``` I can't really think of any features that would be unique to `GRanges` objects since...

For what it's worth, I think the failed CI is a bug in `master`.

I had a similar issue and fixed by installing `tensorflow-gpu==1.8` (assuming you're using a GPU). Conda seems to pull the wrong CuDnn with `v1.7`.

I don't know the answer, but I am curious in general what docs exist about the multitrack features. My assumption is that no additional compression is happening for multitrack files,...

I think I have a related issue, but want to confirm. Say I have the following structure: ``` 1 2 3 ``` Ideally, I'd like to deserialize into the following:...

I guess actually, just using `I()` works: ``` df % mutate(motif = I(ifelse(name == "motif2", motif_rc(edit_motif(motif)), edit_motif(motif)))) ``` Or, of course, calling `update_motif()` after an operation like this.

Yeah, that's a good point. Perhaps the solution is just to do `update_motifs()`. I agree with the sentiment that it gets tedious to have to handle all these special cases,...

Is this needed? If the list is passed to `cmd_args_to_flags()` NA's are dropped. But maybe someone will want to have conditional behavior with NA value? Need to think more about...

This sounds like a useful feature. I've certainly hand-rolled this a few times. Two use-cases come to mind for me: 1. Wanting to flip an entire granges object, like `flip_strand(my_features)`...