|
| 1 | +# Summary of Changes to Metric_computation.py Since Sept 2024 |
| 2 | + |
| 3 | + Timeline of Changes |
| 4 | + |
| 5 | + | Date | Commit | Summary | |
| 6 | + |------------|---------|------------------------------------------| |
| 7 | + | 2025-04-07 | fdc6a0c | Added import_folder() function | |
| 8 | + | 2025-04-11 | bfa002a | Bugfix | |
| 9 | + | 2025-04-14 | e25cb77 | Path updates | |
| 10 | + | 2025-04-29 | 77d8cab | Processing resubmission data | |
| 11 | + | 2025-05-14 | 906998e | Batch processing scripts | |
| 12 | + | 2025-07-30 | 6fc7597 | Restored to using Quilt data | |
| 13 | + | 2025-08-21 | 61904ba | Column name updates (Movie ID → Data ID) | |
| 14 | + | 2025-08-25 | 91db0e9 | Removed unused metric column | |
| 15 | + |
| 16 | + --- |
| 17 | + Major Changes Analysis |
| 18 | + |
| 19 | + 1. IO Changes (Not Breaking) |
| 20 | + |
| 21 | + - Movie ID → Data ID column name changes throughout |
| 22 | + - Restored io.load_bf_colony_features() and io.load_imaging_and_segmentation_dataset() instead of CSV file loading |
| 23 | + - Removed temporary import_folder() function |
| 24 | + |
| 25 | + 2. LOGIC CHANGE: Migration Time Calculation REMOVED ⚠️ |
| 26 | + |
| 27 | + Before (April 2025): |
| 28 | + ``` |
| 29 | + # In add_bottom_mip_migration(): |
| 30 | + raw_values = df_area['Area at the glass (pixels)'].values |
| 31 | + df_area['dy2'] = savgol_filter(raw_values, polyorder=2, window_length=40, deriv=2) |
| 32 | + d_filt = df_area[(df_area.Timepoint>=35) & (df_area.Timepoint<=80)] |
| 33 | + index_infl = d_filt['dy2'].idxmax() |
| 34 | + x_p = df_area['Timepoint'][index_infl] |
| 35 | + df_area['Migration time (h)'] = x_p * (30/60) |
| 36 | + ``` |
| 37 | + |
| 38 | + After (August 2025): This entire migration time calculation was REMOVED. |
| 39 | + |
| 40 | + Impact: The Migration time (h) column is no longer computed in add_bottom_mip_migration(). This is a significant |
| 41 | + logic removal. |
| 42 | + |
| 43 | + 3. LOGIC CHANGE: Gene Metrics Restored ✓ |
| 44 | + |
| 45 | + Before (April 2025): Gene-specific metrics were commented out: |
| 46 | + ``` |
| 47 | + # ######--computing Time of max EOMES expression ------ ##### |
| 48 | + # df_eomes=df_int[(df_int.Gene=='EOMES|TBR2') | (df_int.Gene=='TBR2|EOMES')] |
| 49 | + # ... (all commented) |
| 50 | + ``` |
| 51 | + |
| 52 | + After (July 2025): Gene metrics were restored and expanded: |
| 53 | + - Time of max EOMES expression |
| 54 | + - Time of max TBXT expression (NEW) |
| 55 | + - Time of inflection of E-cad expression |
| 56 | + - Time of half-maximal SOX2 expression |
| 57 | + |
| 58 | + 4. Output Column Removal |
| 59 | + |
| 60 | + Removed from final output (August 2025): |
| 61 | + ``` |
| 62 | + # Removed: 'Time of max expression (h)' |
| 63 | + ``` |
| 64 | + This column was removed from the final feature list. |
| 65 | + |
| 66 | + --- |
| 67 | + #Summary Table |
| 68 | + |
| 69 | + | Change | Type | Impact | |
| 70 | + |------------------------------------------------|--------|------------------------------------------------------------| |
| 71 | + | Movie ID → Data ID | IO | Not breaking | |
| 72 | + | Quilt data loading restored | IO | Not breaking | |
| 73 | + | Migration time calculation REMOVED | LOGIC | Breaking - metric no longer computed | |
| 74 | + | Gene-specific metrics RESTORED | LOGIC | Significant - EOMES, TBXT, CDH1, SOX2 metrics now computed | |
| 75 | + | Time of max expression (h) removed from output | Output | Minor - column removed | |
| 76 | + | import_folder() removed | IO | Not breaking (was temporary) | |
| 77 | + --- |
| 78 | + Conclusion |
| 79 | + |
| 80 | + Two significant logic changes: |
| 81 | + |
| 82 | + 1. Migration time calculation REMOVED from add_bottom_mip_migration() - this metric based on second derivative of |
| 83 | + area at glass is no longer computed |
| 84 | + 2. Gene-specific metrics RESTORED - Time of max EOMES, TBXT expression, E-cad inflection, and SOX2 half-maximal are |
| 85 | + now computed (were commented out before) |
| 86 | + |
| 87 | + The migration time removal could be significant if downstream analysis depended on that metric. |
| 88 | + |
0 commit comments