JOOOOOOOOOOJ commited on
Commit
2a24ff5
·
verified ·
1 Parent(s): 9a23f11

Rename the SLAE-ERR training table (drop v3)

Browse files
.gitattributes CHANGED
@@ -16,3 +16,4 @@ rBMRB/rBMRB-PANAV/test.json filter=lfs diff=lfs merge=lfs -text
16
  rBMRB/rBMRB-PANAV/train.json filter=lfs diff=lfs merge=lfs -text
17
  rBMRB/rBMRB-PANAV/val.json filter=lfs diff=lfs merge=lfs -text
18
  rBMRB/whole_BMRB/train.json filter=lfs diff=lfs merge=lfs -text
 
 
16
  rBMRB/rBMRB-PANAV/train.json filter=lfs diff=lfs merge=lfs -text
17
  rBMRB/rBMRB-PANAV/val.json filter=lfs diff=lfs merge=lfs -text
18
  rBMRB/whole_BMRB/train.json filter=lfs diff=lfs merge=lfs -text
19
+ rBMRB/rBMRB-ERROR/train_val_test_oof_ensemble_10seed_N.json filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -137,7 +137,7 @@ All paths are relative to `rBMRB/`.
137
  | `structures_rBMRB/train_cluster/result_cluster.tsv` | Foldseek clusters of the rBMRB training entries used for out-of-fold (representative, member); 2,606 clusters | 72 kB |
138
  | `structures_whole_BMRB_train/part-00{0..3}.tar.gz`, `index.csv` | AlphaFold2 structures of the whole-BMRB training entries, packed in 4 archives (`<entry_ID>.pdb` at the top level of each); `index.csv` lists which archive holds each file | 12,054 files, 0.36 GB packed (1.6 GB unpacked) |
139
  | `rBMRB-ERROR/{train,val,test}.parquet` | SLAE-ERR training data, one row per rBMRB entry, same split as rBMRB (fields below) | 54 MB in total |
140
- | `rBMRB-ERROR/train_val_test_oof_ensemble_10seed_v3_N.json` | the same data for all splits in one column-oriented json, as read by the SLAE-ERR training code | 175 MB |
141
  | `rBMRB-IDENTITY/identity_pairs.parquet` (+ `.csv`) | identical-sequence deposition pairs used for the intrinsic-noise analysis (Fig. 3), IDs only | 12 kB |
142
  | `rbmrb_entry_conditions.parquet` (+ `.csv`) | sample conditions of all 7,838 curated rBMRB entries (superset of the split) | 0.3 MB |
143
  | `tag_labels_rBMRB.csv` | per-residue tag / linker labels, rBMRB entries | 1.6 MB |
 
137
  | `structures_rBMRB/train_cluster/result_cluster.tsv` | Foldseek clusters of the rBMRB training entries used for out-of-fold (representative, member); 2,606 clusters | 72 kB |
138
  | `structures_whole_BMRB_train/part-00{0..3}.tar.gz`, `index.csv` | AlphaFold2 structures of the whole-BMRB training entries, packed in 4 archives (`<entry_ID>.pdb` at the top level of each); `index.csv` lists which archive holds each file | 12,054 files, 0.36 GB packed (1.6 GB unpacked) |
139
  | `rBMRB-ERROR/{train,val,test}.parquet` | SLAE-ERR training data, one row per rBMRB entry, same split as rBMRB (fields below) | 54 MB in total |
140
+ | `rBMRB-ERROR/train_val_test_oof_ensemble_10seed_N.json` | the same data for all splits in one column-oriented json, as read by the SLAE-ERR training code | 175 MB |
141
  | `rBMRB-IDENTITY/identity_pairs.parquet` (+ `.csv`) | identical-sequence deposition pairs used for the intrinsic-noise analysis (Fig. 3), IDs only | 12 kB |
142
  | `rbmrb_entry_conditions.parquet` (+ `.csv`) | sample conditions of all 7,838 curated rBMRB entries (superset of the split) | 0.3 MB |
143
  | `tag_labels_rBMRB.csv` | per-residue tag / linker labels, rBMRB entries | 1.6 MB |
rBMRB/rBMRB-ERROR/{train_val_test_oof_ensemble_10seed_v3_N.json → train_val_test_oof_ensemble_10seed_N.json} RENAMED
File without changes