"use_priors" is invalid when running ti_paga and ti_paga_tree

[phitlingt]

Hello,

After updating dynwrap to the latest version, ti_paga_tree and ti_paga are not working anymore.
(Problem starts from commit ef9580e, reverting to commit 1ca3ff9 and the error goes away)

I encounter following error while running ti_paga_tree and ti_paga.

library(dyntoy)
toy_data <- dyntoy::generate_dataset(
   id = "specific_example/paga",
   num_cells = 1000,
   num_features = 101,
   model = "tree",
   normalise = FALSE
 )
toy_model <- dynwrap::infer_trajectory(toy_data, ti_paga(n_neighbors=15L, n_comps=50L, 
n_dcs=15L, resolution=1L, embedding_type="umap", connectivity_cutoff=0.05), verbose=TRUE)

Executing 'paga' on 'specific_example/paga'
With parameters: list(n_neighbors = 15L, n_comps = 50L, n_dcs = 15L, resolution = 1L, embedding_type = "umap", connectivity_cutoff = 0.05),
inputs: counts, and
priors : start_id
Input saved to /tmp/folders/x1/nwl1vr796hx3z7fqgwn90mqw0000gn/T//RtmpBi1LRK/file36d832a88530/ti
Running method using babelwhale
Running /usr/local/bin/docker run -e 'TMPDIR=/tmp2' --workdir /ti/workspace -v
'/tmp/folders/x1/nwl1vr796hx3z7fqgwn90mqw0000gn/T//RtmpBi1LRK/file36d832a88530/ti:/ti' -v
'/tmp/folders/x1/nwl1vr796hx3z7fqgwn90mqw0000gn/T//RtmpBi1LRK/file36d855c02c4b/tmp:/tmp2' 'dynverse/ti_paga:v0.9.9.01' --dataset /ti/input.h5
--output /ti/output.h5 --use_priors all
R[write to console]: Error in parse_args(parser, args = args) :
Error in getopt(spec = spec, opt = args) :
long flag "use_priors" is invalid
Calls: -> -> -> parse_args

Running ti_slingshot that uses "--use_priors all" as well was OK with no error.

I checked that ti_paga and ti_paga_tree both have VERSION=0.9.9.03
but my dynwrap::infer_trajectory is calling v0.9.9.01, would this be the problem?
I tried uninstall everything from dynverse and reinstall them back, restarting RStudio and laptop but problem still persists.

[rcannood]

Hello Phit Ling!

Could you reinstall dynmethods? This problem had been fixed in the devel branch of the package, I just forgot to merge it to master.

Relatedly, I noticed that the umap argument of paga does not seem to be working. I will create a separate issue for this.

If you reinstall dynmethods and remove the embedding type parameter, can you verify it works for you now?

Robrecht

[rcannood]

FYI, if you want to try out different versions of docker containers, or custom docker containers, you can use the following code:

my_method <- dynwrap::create_ti_method_container("dynverse/ti_paga:v0.9.9.03")
toy_model <- dynwrap::infer_trajectory(
  toy_data, 
  my_method(n_neighbors=15L),
  verbose=TRUE
)

[phitlingt]

It works!
Reinstalled dynmethods and removed the embedding type parameter did the job.

Thanks for the great tip regarding the docker containers too.

Really appreciate your team's prompt reply and wonderful work!