ORFs in the same well position

jump-cellpainting · Oct 25, 2023 · 6e7b008 · 6e7b008
1 parent 9b066e0
commit 6e7b008
Show file tree

Hide file tree

Showing 11 changed files with 1,857 additions and 10 deletions.
diff --git a/benchmark/1.calculate-map-cp.ipynb → benchmark/1.0.calculate-map-cp.ipynb b/benchmark/1.calculate-map-cp.ipynb → benchmark/1.0.calculate-map-cp.ipynb
@@ -95,13 +95,13 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "Normalized, feature selected compound profiles for replicate plates at the same time point, using the same cell type are merged into a single data frame. Empty wells are removed before calculating mAP for perturbation detection (retrieve replicates of the query compound). mAP for the random baseline (10000 random list of size number of replicates + number of negcon) is subtracted.\n",
+    "Normalized, feature selected compound profiles for replicate plates at the same time point, using the same cell type are merged into a single data frame. Empty wells are removed before calculating mAP for perturbation detection (retrieve replicates of the query compound). p value and FDR corrected p values are calculated using a permutation-based significance test.\n",
     "\n",
-    "For the perturbation matching task, compounds that not are replicable (qvalue > 0.05) are removed. Negative control wells are also removed and the additional target annotations are added before mAP for perturbation matching is calculated. Again, the mAP for the random baseline is subtracted.\n",
+    "For the perturbation matching task, compounds that not are replicable (qvalue > 0.05) are removed. Negative control wells are also removed and the additional target annotations are added before mAP for perturbation matching is calculated. Again, p value and FDR corrected p values are calculated using a permutation-based significance test.\n",
     "\n",
     "Similarly, mAP for perturbation detection is computed for both CRISPR and ORF plates while mAP for perturbation matching is computed only for the CRISPR guides as there are no sister ORF perturbations.\n",
     "\n",
-    "Then, mAP for the chemical perturbation-genetic perturbation matching task is computed using the median consensus profiles and the baseline mAP is subtracted.\n",
+    "Then, mAP for the chemical perturbation-genetic perturbation matching task is computed using the median consensus profiles.\n",
     "\n",
     "Along with mAP, fraction retrieved is also calculated. This is the fraction of queries that have qvalue < 0.05.\n",
     "\n",

diff --git a/benchmark/1.1.calculate-map-cp-orf.ipynb b/benchmark/1.1.calculate-map-cp-orf.ipynb
diff --git a/benchmark/2.generate-map-figure.ipynb b/benchmark/2.generate-map-figure.ipynb
diff --git a/benchmark/figures/1.1.CellProfiler_replicability_orf_fr.png b/benchmark/figures/1.1.CellProfiler_replicability_orf_fr.png
diff --git a/benchmark/figures/1.1.CellProfiler_replicability_orf_map.png b/benchmark/figures/1.1.CellProfiler_replicability_orf_map.png
diff --git a/benchmark/figures/2.gene_compound_matching_map.png b/benchmark/figures/2.gene_compound_matching_map.png
diff --git a/benchmark/figures/2.matching_map.png b/benchmark/figures/2.matching_map.png
diff --git a/benchmark/figures/2.replicability_map.png b/benchmark/figures/2.replicability_map.png
diff --git a/benchmark/figures/2.replicability_map_with_legend.png b/benchmark/figures/2.replicability_map_with_legend.png
diff --git a/benchmark/output/cellprofiler_replicability_orf_fr.csv b/benchmark/output/cellprofiler_replicability_orf_fr.csv
@@ -0,0 +1,5 @@
+Description,Modality,Cell,time,timepoint,fr
+orf_A549_long,orf,A549,long,96,0.247
+orf_A549_short,orf,A549,short,48,0.234
+orf_U2OS_short,orf,U2OS,short,48,0.425
+orf_U2OS_long,orf,U2OS,long,96,0.241
diff --git a/benchmark/output/cellprofiler_replicability_orf_map.csv b/benchmark/output/cellprofiler_replicability_orf_map.csv