Skip "assign_gRNA" step in local laptop

[nhu-github]

Hi, as I went through of "assign_gRNA" step, I did not find any parameter setup to skip it even thought I noticed it is an optional step. This is in local laptop. Any suggestion on how to skip this step as my data already assigned gRNAs. https://timothy-barry.github.io/sceptre-book/assign-grnas.html

This try does not work though:
sceptre_object@last_function_called <- as.character('assign_grnas')

Error in perform_status_check_and_update(sceptre_object, "run_qc") : The function `assign_grnas()` must be called before `run_qc()` is called

[timothy-barry]

Hi Nan,

"Optional" in this context means that sceptre automatically calls assign_grnas() (with default parameters) if the user does not call this function.

Would you mind sharing a bit more detail about your analysis? In particular, did you import the data via import_data()? If so, what did you supply to the grna_matrix argument? A binary matrix of gRNA-to-cell assignments? If so, my recommendation would be to explicitly call assign_grnas(), setting method to "thresholding" and threshold to 1. This will cause sceptre to use your custom assignments.

[nhu-github]

Thanks for your reply. So I follow the standard pipeline from the beginning to the end of discovery analysis and save the final sceptre object. This time I re-run set-analysis-parameter() to change the discovery_pairs to the previous sceptre object which means the gRNA is already assigned in previous analysis. Then I would like to skip assign_gRNA(), but the program did not allow me to. thresholding=1 will change the original gRNA assignment.

[timothy-barry]

Oh, I see. So this is a question about caching/optimization. sceptre currently supports basic caching, but it is not super sophisticated. We plan to add more sophisticated caching functionality in a future update.

For the time being, I would recommend calling sceptre_object <- sceptre:::perform_status_check_and_update(sceptre_object, "assign_grnas") instead of assign_grnas(), as follows.

library(sceptredata)
library(sceptre)

directories <- paste0(
  system.file("extdata", package = "sceptredata"),
  "/highmoi_example/gem_group_", c(1, 2)
)
data(grna_target_data_frame_highmoi)
sceptre_object <- import_data_from_cellranger(
  directories = directories,
  moi = "high",
  grna_target_data_frame = grna_target_data_frame_highmoi
)
print(sceptre_object)

# 2. set the analysis parameters
positive_control_pairs <- construct_positive_control_pairs(sceptre_object)
discovery_pairs <- construct_cis_pairs(sceptre_object,
                                       positive_control_pairs = positive_control_pairs,
                                       distance_threshold = 5e6
)

sceptre_object <- set_analysis_parameters(
  sceptre_object = sceptre_object,
  discovery_pairs = discovery_pairs,
  positive_control_pairs = positive_control_pairs,
  side = "left"
)
print(sceptre_object)

# 3. assign grnas
plot_grna_count_distributions(sceptre_object)
sceptre_object <- sceptre_object |> assign_grnas(parallel = TRUE, n_processors = 2)
plot(sceptre_object)
print(sceptre_object)

# 4. run qc
plot_covariates(sceptre_object, p_mito_threshold = 0.075)
sceptre_object <- sceptre_object |> run_qc(p_mito_threshold = 0.075)
plot(sceptre_object)
print(sceptre_object)

# 5. run the calibration check
sceptre_object <- run_calibration_check(sceptre_object, parallel = TRUE, n_processors = 2)
plot(sceptre_object)
print(sceptre_object)

# 6. run the power check
sceptre_object <- run_power_check(sceptre_object, parallel = TRUE, n_processors = 2)
plot(sceptre_object)
print(sceptre_object)

# 7. run discovery analysis
sceptre_object <- run_discovery_analysis(sceptre_object, parallel = TRUE, n_processors = 2)
plot(sceptre_object)
print(sceptre_object)

# 8. reassign the discovery pairs
discovery_pairs <- construct_cis_pairs(sceptre_object,
                                       positive_control_pairs = positive_control_pairs,
                                       distance_threshold = 1e6
)
sceptre_object <- set_analysis_parameters(
  sceptre_object = sceptre_object,
  discovery_pairs = discovery_pairs,
  positive_control_pairs = positive_control_pairs,
  side = "left"
)

# 9. skip the grna assignment step
sceptre_object <- sceptre:::perform_status_check_and_update(sceptre_object, "assign_grnas")

# 10. proceed with the rest of  theanalysis
sceptre_object <- sceptre_object |> run_qc() 
sceptre_object <- run_calibration_check(sceptre_object, parallel = TRUE, n_processors = 2) # etc

[nhu-github]

Thanks. This works.