try add tests

.github/workflows/CI.yml

@@ -20,6 +20,8 @@ jobs:
           - "ukb_allele_independent.jl"
           - "custom_from_actors.jl"
           - "make_dataset.jl"
+          - "null_simulation.jl"
+          - "realistic_simulation.jl"
     steps:
       - uses: actions/checkout@v2
       - uses: julia-actions/setup-julia@v1

README.md

@@ -21,4 +21,4 @@ nextflow run TARGENE/targene-pipeline -profile myprofile -resume
 ```julia
 using LiveServer
 servedocs()
-```
+```

conf/container.config

@@ -1,6 +1,6 @@
 process {
     withLabel: simulation_image {
-        container = "olivierlabayle/targene-simulations:main"
+        container = "olivierlabayle/targene-simulations:ps_simulation"
     }
     withLabel: targenecore_image {
         container = "olivierlabayle/tl-core:simulation"
@@ -9,7 +9,7 @@ process {
         container = "olivierlabayle/ukbmain:0.4"
     }
     withLabel: tmle_image {
-        container = "olivierlabayle/targeted-estimation:0.8"
+        container = "olivierlabayle/targeted-estimation:agnostic_composed"
     }
     withLabel: plink_image {
         container = "olivierlabayle/plink2:0.1.0"

main.nf

@@ -13,6 +13,8 @@ params.COHORT = "UKBB"
 params.UKB_CONFIG = "${projectDir}/assets/ukbconfig.yaml"
 params.UKB_WITHDRAWAL_LIST = "${projectDir}/assets/NO_WITHDRAWAL_LIST"
 params.OUTDIR = "${launchDir}/results"
+params.VARIANTS_REGEXP = "^(rs[0-9]*|Affx)"
+params.SAMPLE_GA_HITS = true
 
 // Confounding adjustment by PCA
 params.NB_PCS = 6
@@ -78,7 +80,7 @@ include { TARGENE } from './workflows/targene.nf'
 include { PERMUTATION_TEST; RANDOMIZATION_TEST } from './workflows/negative_control.nf'
 include { PCA } from './workflows/pca.nf'
 include { MAKE_DATASET } from './workflows/dataset.nf'
-include { NULL_SIMULATION; REALISTIC_SIMULATION; SIMULATIONS } from './workflows/simulations.nf'
+include { NULL_SIMULATION; REALISTIC_SIMULATION } from './workflows/simulations.nf'
 
 log.info """\
          ${workflow.manifest.name} v${workflow.manifest.version}

modules/simulations.nf

@@ -110,6 +110,7 @@ process RealisticSimulationInputs {
         TEMPD=\$(mktemp -d)
         JULIA_DEPOT_PATH=\$TEMPD:/opt julia --project=/opt/Simulations --startup-file=no --sysimage=/opt/Simulations/Simulations.so /opt/Simulations/targene-simulation.jl \
         simulation-inputs-from-ga ${estimands_prefix} ${bgen_prefix} ${traits} ${pcs} \
+        --sample-gene-atlas-hits=${params.SAMPLE_GA_HITS} \
         --ga-download-dir=gene_atlas \
         --remove-ga-data=true \
         --ga-trait-table=${ga_trait_table} \
@@ -121,6 +122,7 @@ process RealisticSimulationInputs {
         --output-prefix=ga_sim_input \
         --batchsize=${params.BATCH_SIZE} \
         --verbosity=${params.VERBOSITY} \
+        --variants-regex=${params.VARIANTS_REGEXP} \
         ${call_threshold}
         """
 }

test/assets/estimators-configs/cv-glm.jl

@@ -0,0 +1,16 @@
+resampling = JointStratifiedCV(patterns=[r"^rs[0-9]+"], resampling=StratifiedCV(nfolds=3))
+
+default_models = TMLE.default_models(
+  # For the estimation of E[Y|W, T]: continuous outcome
+  Q_continuous = LinearRegressor(),
+  # For the estimation of E[Y|W, T]: binary target
+  Q_binary = LogisticClassifier(lambda=0.),
+  # For the estimation of p(T| W)
+  G = LogisticClassifier(lambda=0.)
+)
+
+ESTIMATORS = (
+  CV_wTMLE_GLMNET = TMLEE(models=default_models, weighted=true, resampling=resampling),
+  CV_TMLE_GLMNET = TMLEE(models=default_models, weighted=false, resampling=resampling),
+  CV_OSE_GLMNET = OSE(models=default_models, resampling=resampling)
+)

test/assets/estimators-configs/glm.jl

@@ -0,0 +1,14 @@
+default_models = TMLE.default_models(
+  # For the estimation of E[Y|W, T]: continuous outcome
+  Q_continuous = LinearRegressor(),
+  # For the estimation of E[Y|W, T]: binary target
+  Q_binary = LogisticClassifier(lambda=0.),
+  # For the estimation of p(T| W)
+  G = LogisticClassifier(lambda=0.)
+)
+
+ESTIMATORS = (
+  wTMLE_GLMNET = TMLEE(models=default_models, weighted=true),
+  TMLE_GLMNET = TMLEE(models=default_models, weighted=false),
+  OSE_GLMNET = OSE(models=default_models)
+)

test/configs/simulation.config

@@ -0,0 +1,17 @@
+includeConfig 'ukb.config'
+
+params {
+    ESTIMANDS_FILE = "test/assets/parameters/parameters.yaml" 
+    SAMPLE_GA_HITS = false // To complicated to test otherwise
+
+    // Grid
+    N_REPEATS = 3
+    RNGS = [0, 1, 3]
+    SAMPLE_SIZES = [1000]
+    ESTIMATORS = "test/assets/estimators-configs/*"
+
+    // Misc
+    POSITIVITY_CONSTRAINT = 0
+    BATCH_SIZE = 3
+    CALL_THRESHOLD = null // This is essential not to have too many missings
+}

test/null_simulation.jl

@@ -0,0 +1,13 @@
+module TestNullSimulation
+
+args = length(ARGS) > 0 ? ARGS : ["-profile", "local", "-resume"] 
+
+@testset "Test Null Simulation Workflow" begin
+    cmd = `nextflow run main.nf -entry NULL_SIMULATION -c test/configs/simulation.config $args`
+    @info string("The following command will be run:\n", cmd)
+
+    r = run(cmd)
+    @test r.exitcode == 0
+end
+
+end

test/realistic_simulation.jl

@@ -0,0 +1,13 @@
+module TestRealisticSimulation
+
+args = length(ARGS) > 0 ? ARGS : ["-profile", "local", "-resume"] 
+
+@testset "Test Realistic Simulation Workflow" begin
+    cmd = `nextflow run main.nf -entry REALISTIC_SIMULATION -c test/configs/simulation.config $args`
+    @info string("The following command will be run:\n", cmd)
+
+    r = run(cmd)
+    @test r.exitcode == 0
+end
+
+end

workflows/simulations.nf

@@ -90,9 +90,4 @@ workflow REALISTIC_SIMULATION {
 
     // Aggregation of Estimation Results
     AggregateSimulationResults(simulation_results.collect(), "realistic_simulation_results.hdf5")
-}
-
-workflow SIMULATIONS {
-    NULL_SIMULATION()
-    REALISTIC_SIMULATION()
 }