Fix #864 - QTL Pipeline Updates #873

jwildfire · 2022-11-10T16:05:39Z

Overview

Updates to QTL pipelines for v1.3.1 release.

Test Notes/Sample Code

Notes:

mattroumaya · 2022-11-11T14:31:22Z

@jwildfire this one just needs unit tests which I'm working on now - feel free to review if you have the chance. I'll convert to PR and approve once everything looks good.

mattroumaya · 2022-11-11T16:00:38Z

R/Flag_QTL.R

+  dfFlagged <- dfAnalyzed %>%
+    mutate(
+      Flag = case_when(
+        (.data$LowCI > vThreshold) ~ 2,
+        (.data$Score > vThreshold) ~ 1,
+        TRUE ~ 0
+      )
+    )


Should we add some NA handling here?

Either stopifnot() for is.na(Score) or handle within case_when()?

dfAnalyzed <- tibble::tribble( ~GroupID, ~Numerator, ~Denominator, ~Metric, ~Method, ~ConfLevel, ~Estimate, ~LowCI, ~UpCI, ~Score, "AA-AA-000-0000", 122, 1301, 0.0937, "Exact binomial test", 0.95, 0.0937, 0.0784, 0.1109, 0.0784 ) # returns Flag == 0 dfAnalyzed %>% mutate(Score = NA) %>% Flag_QTL(vThreshold = 1) # returns Flag == 2 dfAnalyzed %>% mutate(Score = NA) %>% Flag_QTL(vThreshold = 0)

Probably not needed. I think as long as denominator >=1, there won't be NA in Estimate, Score or LowCI.

Also, need to update Score in Analyze_QTL() to be the Estimate for flagging 1. LowerCI is used for flagging 2.

Merge remote-tracking branch 'origin/dev' into fix-864 # Conflicts: # R/Disp_Assess.R # R/LB_Assess.R # R/PD_Assess.R # man/Disp_Assess.Rd # man/LB_Assess.Rd # man/PD_Assess.Rd

R/Flag_QTL.R

gwu05 · 2022-11-14T23:35:37Z

R/Disp_Assess.R

    }

    flag_function_name <- switch(strMethod,
      NormalApprox = "Flag_NormalApprox",
      identity = "Flag",
      fisher = "Flag_Fisher",
-      qtl = "Flag"
+      qtl = "Flag_QTL"


is there a qtl strMethod? this sections differs between AE and Disp

We're only running QTLs for Disp and PD right now - should we add the QTL method to AE + others?

oh you're right, that's correct

gwu05 · 2022-11-14T23:44:23Z

R/Make_Snapshot.R

+          select(GroupID,
+                 LowCI,
+                 UpCI,
+                 Score) %>%


Not sure if we need to pass through more - eg., whether the datapoint was flagged and the corresponding threshold - in order to allow for changes to QTL thresholds through the study. As long as we can get this elsewhere then we are fine. Also, given that we are only using an upper threshold for QTLs currently, UpCI actually technically doesn't need to be surfaced (but okay to include for now). In general, it'd probably be better to allow for lower and upper QTL thresholds in our Flag_QTL function (although I don't see us using the lower QTL in the recent release, so it can be something saved for later.)

gwu05 · 2022-11-14T23:45:57Z

R/Make_Snapshot.R

@@ -200,10 +200,40 @@ bQuiet = TRUE
      flag = "Flag"
    )



I was not able to run the example in Make_Snapshot:

"Error in select():
! Can't subset columns that don't exist.
✖ Column lpfv doesn't exist.
Run rlang::last_error() to see where the error occurred."

Backtrace:
▆

├─gsm::Make_Snapshot()

│ └─status_study %>% ... at gsm/R/Make_Snapshot.R:93:2

├─dplyr::select(...)

└─dplyr:::select.data.frame(...)
...

I think you might need to update to the latest {clindata} release because lpfv was recently added

R/PD_Assess.R

jwildfire

Looks good. I approve (but I can't officially approve, since I started the PR).

@mattroumaya I think you're good to approve/merge.

jwildfire · 2022-11-15T16:28:46Z

R/Make_Snapshot.R

+  # Also need to make sure we're capturing WorkflowID here ...
+
+  hasQTL <- grep("qtl", names(lResults))
+


A+ data munging. 💯

mattroumaya · 2022-11-15T17:37:04Z

R/Make_Snapshot.R

+  } else {
+    results_analysis <- tibble(
+      studyid = NA,
+      workflowid = NA,


add workflowid and studyid if QTL fails

mattroumaya · 2022-11-15T18:42:28Z

@jwildfire - had to tweak the logic a little bit for a failed QTL, and also updated rbm_data_spec. Let me know if this all looks good and then I'll merge!

mattroumaya · 2022-11-15T19:10:31Z

Also added in failsafe logic for the case when none of the workflows contain dfBounds:

lAssessments <- MakeWorkflowList(strNames = c("kri0001", "qtl0003", "qtl0007"))

lAssessments$kri0001$steps[[2]]$inputs <- "dfAA"

a <- Make_Snapshot(lAssessments = lAssessments)

jwildfire

Updates look good

jwildfire added 3 commits November 10, 2022 09:53

remove strType. fix #849

1a504d2

prototype for QTL flag method. #864

91d3636

spec for results_analysis. #863

1385958

jwildfire requested review from mattroumaya and Zhongkai-Wang November 10, 2022 16:05

jwildfire assigned mattroumaya Nov 10, 2022

jwildfire marked this pull request as draft November 10, 2022 16:05

add results_analysis + cleanup

4f21bb2

mattroumaya reviewed Nov 11, 2022

View reviewed changes

mattroumaya added 4 commits November 11, 2022 16:08

add unit tests

9035715

add logic to check if qtl exists in lResults

b7e52b2

change score to estimate

da60476

ffwd dev

ef46a5f

Merge remote-tracking branch 'origin/dev' into fix-864 # Conflicts: # R/Disp_Assess.R # R/LB_Assess.R # R/PD_Assess.R # man/Disp_Assess.Rd # man/LB_Assess.Rd # man/PD_Assess.Rd

mattroumaya marked this pull request as ready for review November 14, 2022 16:34

mattroumaya requested a review from gwu05 November 14, 2022 16:46

gwu05 reviewed Nov 14, 2022

View reviewed changes

R/Flag_QTL.R Show resolved Hide resolved

gwu05 reviewed Nov 14, 2022

View reviewed changes

R/PD_Assess.R Outdated Show resolved Hide resolved

review updates

812184a

jwildfire commented Nov 15, 2022

View reviewed changes

gwu05 approved these changes Nov 15, 2022

View reviewed changes

Merge remote-tracking branch 'origin/dev' into fix-864

8b558be

mattroumaya reviewed Nov 15, 2022

View reviewed changes

mattroumaya added 2 commits November 15, 2022 18:25

tweak logic for failed QTL

1dfacc1

update rbm_data_spec

4199683

add failsafe for dfBounds

bc1305a

jwildfire commented Nov 15, 2022

View reviewed changes

mattroumaya merged commit 2149c90 into dev Nov 15, 2022

mattroumaya deleted the fix-864 branch November 15, 2022 19:46

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix #864 - QTL Pipeline Updates #873

Fix #864 - QTL Pipeline Updates #873

jwildfire commented Nov 10, 2022 •

edited by mattroumaya

Loading

mattroumaya commented Nov 11, 2022

mattroumaya Nov 11, 2022

Zhongkai-Wang Nov 11, 2022 •

edited

Loading

gwu05 Nov 14, 2022

mattroumaya Nov 15, 2022

gwu05 Nov 15, 2022

gwu05 Nov 14, 2022 •

edited

Loading

gwu05 Nov 14, 2022 •

edited

Loading

mattroumaya Nov 15, 2022

jwildfire left a comment

jwildfire Nov 15, 2022

mattroumaya Nov 15, 2022

mattroumaya commented Nov 15, 2022

mattroumaya commented Nov 15, 2022

jwildfire left a comment

		# Also need to make sure we're capturing WorkflowID here ...

		hasQTL <- grep("qtl", names(lResults))

Fix #864 - QTL Pipeline Updates #873

Fix #864 - QTL Pipeline Updates #873

Conversation

jwildfire commented Nov 10, 2022 • edited by mattroumaya Loading

Overview

Test Notes/Sample Code

mattroumaya commented Nov 11, 2022

mattroumaya Nov 11, 2022

Choose a reason for hiding this comment

Zhongkai-Wang Nov 11, 2022 • edited Loading

Choose a reason for hiding this comment

gwu05 Nov 14, 2022

Choose a reason for hiding this comment

mattroumaya Nov 15, 2022

Choose a reason for hiding this comment

gwu05 Nov 15, 2022

Choose a reason for hiding this comment

gwu05 Nov 14, 2022 • edited Loading

Choose a reason for hiding this comment

gwu05 Nov 14, 2022 • edited Loading

Choose a reason for hiding this comment

"Error in select(): ! Can't subset columns that don't exist. ✖ Column lpfv doesn't exist. Run rlang::last_error() to see where the error occurred."

mattroumaya Nov 15, 2022

Choose a reason for hiding this comment

jwildfire left a comment

Choose a reason for hiding this comment

jwildfire Nov 15, 2022

Choose a reason for hiding this comment

mattroumaya Nov 15, 2022

Choose a reason for hiding this comment

mattroumaya commented Nov 15, 2022

mattroumaya commented Nov 15, 2022

jwildfire left a comment

Choose a reason for hiding this comment

jwildfire commented Nov 10, 2022 •

edited by mattroumaya

Loading

Zhongkai-Wang Nov 11, 2022 •

edited

Loading

gwu05 Nov 14, 2022 •

edited

Loading

gwu05 Nov 14, 2022 •

edited

Loading

"Error in `select()`:
! Can't subset columns that don't exist.
✖ Column `lpfv` doesn't exist.
Run `rlang::last_error()` to see where the error occurred."