Return unused deepclust70 held-out reps to train; drop test_extra; card cleanup
Browse files- README.md +31 -40
- deepclust70/returned.json +40 -0
- deepclust70/test_extra_gte2.parquet +0 -3
- deepclust70/test_extra_singletons.parquet +0 -3
- deepclust70/train/shard-05353.parquet +2 -2
- manifest.parquet +2 -2
README.md
CHANGED
|
@@ -24,8 +24,8 @@ protein language model. Two sources, mixed by rows at training time:
|
|
| 24 |
sequences clustered with DIAMOND v2.2.5 `linclust` (`--approx-id 70 --member-cover 90`).
|
| 25 |
One representative per row, globally shuffled. Re-cut to 1,000,000-row shards by `oplm data reshard`.
|
| 26 |
|
| 27 |
-
Every number below is copied from `uniref70/summary.json`, `deepclust70/
|
| 28 |
-
`deepclust70/
|
| 29 |
[`docs/data.md`](https://github.com/briney/oplm/blob/main/docs/data.md).
|
| 30 |
|
| 31 |
## Layout
|
|
@@ -42,14 +42,13 @@ uniref70/
|
|
| 42 |
test_gte2.parquet # 5,000
|
| 43 |
deepclust70/
|
| 44 |
train/summary.json # re-cut record
|
| 45 |
-
train/shard-00000.parquet # 5,354 shards, 1,000,000 rows each (the last has
|
| 46 |
-
heldout.json #
|
|
|
|
| 47 |
eval_singletons.parquet # 5,000 held-out singleton-cluster reps
|
| 48 |
eval_gte2.parquet # 5,000 held-out reps of 2+-member clusters
|
| 49 |
test_singletons.parquet # 5,000
|
| 50 |
test_gte2.parquet # 5,000
|
| 51 |
-
test_extra_singletons.parquet # 105,279 (every other screen survivor)
|
| 52 |
-
test_extra_gte2.parquet # 79,684
|
| 53 |
```
|
| 54 |
|
| 55 |
`manifest.parquet` lists every file with its row count, size and sha256. Check downloads
|
|
@@ -98,56 +97,35 @@ Evaluation uses the representatives in both phases.
|
|
| 98 |
| `uniref70/eval_gte2.parquet` | 5,000 | 11,119 | 1,973,383 | 4,469,677 |
|
| 99 |
| `uniref70/test_singletons.parquet` | 5,000 | 5,000 | 1,701,713 | 1,701,713 |
|
| 100 |
| `uniref70/test_gte2.parquet` | 5,000 | 11,204 | 1,891,172 | 4,295,018 |
|
| 101 |
-
| `deepclust70/train/` | 5,353,
|
| 102 |
| `deepclust70/eval_singletons.parquet` | 5,000 | | 486,576 | |
|
| 103 |
| `deepclust70/eval_gte2.parquet` | 5,000 | | 887,835 | |
|
| 104 |
| `deepclust70/test_singletons.parquet` | 5,000 | | 488,725 | |
|
| 105 |
| `deepclust70/test_gte2.parquet` | 5,000 | | 877,340 | |
|
| 106 |
-
| `deepclust70/test_extra_singletons.parquet` | 105,279 | | 10,377,799 | |
|
| 107 |
-
| `deepclust70/test_extra_gte2.parquet` | 79,684 | | 14,222,233 | |
|
| 108 |
|
| 109 |
uniref70: 121,494,192 UniRef90 sequences in, 75,738,236 clusters out. Members per train
|
| 110 |
cluster: min 1, median 1, mean 1.604, max 2,370; 60,298,749 train clusters are
|
| 111 |
singletons.
|
| 112 |
|
| 113 |
-
##
|
| 114 |
|
| 115 |
-
Eval and test
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
|
| 120 |
```
|
| 121 |
--min-seq-id 0.7 -c 0.8 --cov-mode 0 -s 1 --max-seqs 100000 -e 1e6
|
| 122 |
```
|
| 123 |
|
| 124 |
-
|
| 125 |
-
|
|
|
|
|
|
|
| 126 |
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
| gte2 | 250,000 | 10,832 | 4.3% |
|
| 131 |
-
|
| 132 |
-
**Caveat.** The gte2 eval and test sets come from the 4.3% of multi-member clusters
|
| 133 |
-
with no ≥ 70%-identity relative anywhere in training. They are a narrower slice than a
|
| 134 |
-
random gte2 draw. The build's 5% stop rule tripped on gte2; the maintainers chose to use
|
| 135 |
-
the survivors anyway (`--min-survival 0.04`). A bigger pool would give more survivors
|
| 136 |
-
at the same rate, because the pool is a uniform draw from the stratum.
|
| 137 |
-
|
| 138 |
-
## deepclust70 eval and test
|
| 139 |
-
|
| 140 |
-
deepclust70's original held-out representatives (random draws from singleton clusters
|
| 141 |
-
and from clusters with 2+ members; 15,000 eval and 150,000 test per stratum) were
|
| 142 |
-
screened with the same MMseqs2 arguments against deepclust70 `train/` and every
|
| 143 |
-
uniref70 `train/` member. Per stratum the pool is the old eval file then the old test
|
| 144 |
-
file. The first 5,000 survivors are eval, the next 5,000 test, and the rest
|
| 145 |
-
`test_extra_*`. The record is `deepclust70/heldout.json`.
|
| 146 |
-
|
| 147 |
-
| stratum | candidates | survivors | survival |
|
| 148 |
-
|---|--:|--:|--:|
|
| 149 |
-
| singletons | 165,000 | 115,279 | 69.9% |
|
| 150 |
-
| gte2 | 165,000 | 89,684 | 54.4% |
|
| 151 |
|
| 152 |
## Rebuild
|
| 153 |
|
|
@@ -178,6 +156,17 @@ had already been run on its own, with the same `easy-linclust` arguments plus
|
|
| 178 |
`--split-memory-limit 80G`, so the run above picked up its output. To rebuild in one
|
| 179 |
command, add `--mmseqs-extra=--split-memory-limit --mmseqs-extra=80G`.
|
| 180 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 181 |
Then write the manifest:
|
| 182 |
|
| 183 |
```
|
|
@@ -189,6 +178,8 @@ oplm data manifest <out>
|
|
| 189 |
uniref70 was built with oplm 0.4.1.dev252+ga400dfede. Clustering ran at commit `ea4b33a`
|
| 190 |
(2026-10-30), the eval screen at `e6f331c` and the write at `a400dfe` (2026-11-01).
|
| 191 |
The clustering stage's `summary.json` entry was written by hand from that run's record.
|
|
|
|
|
|
|
| 192 |
|
| 193 |
## License and attribution
|
| 194 |
|
|
|
|
| 24 |
sequences clustered with DIAMOND v2.2.5 `linclust` (`--approx-id 70 --member-cover 90`).
|
| 25 |
One representative per row, globally shuffled. Re-cut to 1,000,000-row shards by `oplm data reshard`.
|
| 26 |
|
| 27 |
+
Every number below is copied from `uniref70/summary.json`, `deepclust70/train/summary.json`
|
| 28 |
+
or `deepclust70/returned.json`. The full provenance is in OPLM's
|
| 29 |
[`docs/data.md`](https://github.com/briney/oplm/blob/main/docs/data.md).
|
| 30 |
|
| 31 |
## Layout
|
|
|
|
| 42 |
test_gte2.parquet # 5,000
|
| 43 |
deepclust70/
|
| 44 |
train/summary.json # re-cut record
|
| 45 |
+
train/shard-00000.parquet # 5,354 shards, 1,000,000 rows each (the last has 896,515)
|
| 46 |
+
heldout.json # eval/test screen record
|
| 47 |
+
returned.json # record of unused held-out reps returned to train/
|
| 48 |
eval_singletons.parquet # 5,000 held-out singleton-cluster reps
|
| 49 |
eval_gte2.parquet # 5,000 held-out reps of 2+-member clusters
|
| 50 |
test_singletons.parquet # 5,000
|
| 51 |
test_gte2.parquet # 5,000
|
|
|
|
|
|
|
| 52 |
```
|
| 53 |
|
| 54 |
`manifest.parquet` lists every file with its row count, size and sha256. Check downloads
|
|
|
|
| 97 |
| `uniref70/eval_gte2.parquet` | 5,000 | 11,119 | 1,973,383 | 4,469,677 |
|
| 98 |
| `uniref70/test_singletons.parquet` | 5,000 | 5,000 | 1,701,713 | 1,701,713 |
|
| 99 |
| `uniref70/test_gte2.parquet` | 5,000 | 11,204 | 1,891,172 | 4,295,018 |
|
| 100 |
+
| `deepclust70/train/` | 5,353,896,515 | | | |
|
| 101 |
| `deepclust70/eval_singletons.parquet` | 5,000 | | 486,576 | |
|
| 102 |
| `deepclust70/eval_gte2.parquet` | 5,000 | | 887,835 | |
|
| 103 |
| `deepclust70/test_singletons.parquet` | 5,000 | | 488,725 | |
|
| 104 |
| `deepclust70/test_gte2.parquet` | 5,000 | | 877,340 | |
|
|
|
|
|
|
|
| 105 |
|
| 106 |
uniref70: 121,494,192 UniRef90 sequences in, 75,738,236 clusters out. Members per train
|
| 107 |
cluster: min 1, median 1, mean 1.604, max 2,370; 60,298,749 train clusters are
|
| 108 |
singletons.
|
| 109 |
|
| 110 |
+
## Eval and test
|
| 111 |
|
| 112 |
+
Eval and test sets are held out of the whole training mix, not just their own source.
|
| 113 |
+
Each stratum is a random draw of that source's clusters: singletons, or clusters with 2+
|
| 114 |
+
members. Every eval/test sequence is searched with MMseqs2 against deepclust70 `train/`
|
| 115 |
+
and against every member of every uniref70 train cluster:
|
| 116 |
|
| 117 |
```
|
| 118 |
--min-seq-id 0.7 -c 0.8 --cov-mode 0 -s 1 --max-seqs 100000 -e 1e6
|
| 119 |
```
|
| 120 |
|
| 121 |
+
Any hit excludes the example. For uniref70, every member of a held-out cluster is
|
| 122 |
+
searched, not just the representative. Searching every uniref70 member, not just the
|
| 123 |
+
representatives, keeps eval and test independent in phase 2, when training samples
|
| 124 |
+
members.
|
| 125 |
|
| 126 |
+
Held-out candidates not used for eval or test are in `train/`, unless they match an
|
| 127 |
+
eval/test sequence under the same search. deepclust70's are shuffled into the last
|
| 128 |
+
`train/` shard.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 129 |
|
| 130 |
## Rebuild
|
| 131 |
|
|
|
|
| 156 |
`--split-memory-limit 80G`, so the run above picked up its output. To rebuild in one
|
| 157 |
command, add `--mmseqs-extra=--split-memory-limit --mmseqs-extra=80G`.
|
| 158 |
|
| 159 |
+
deepclust70 eval/test, then the unused held-out reps back to `train/`
|
| 160 |
+
(`<held-out dir>` holds deepclust70's original held-out representatives):
|
| 161 |
+
|
| 162 |
+
```
|
| 163 |
+
oplm data deepclust70-heldout <held-out dir> <out>/deepclust70 \
|
| 164 |
+
--deepclust70 <out>/deepclust70/train --uniref70 <out>/uniref70/train \
|
| 165 |
+
--workdir <work> --threads 32
|
| 166 |
+
oplm data deepclust70-return <held-out dir> <out>/deepclust70 \
|
| 167 |
+
--uniref70 <out>/uniref70 --workdir <work2> --threads 32 --seed 0
|
| 168 |
+
```
|
| 169 |
+
|
| 170 |
Then write the manifest:
|
| 171 |
|
| 172 |
```
|
|
|
|
| 178 |
uniref70 was built with oplm 0.4.1.dev252+ga400dfede. Clustering ran at commit `ea4b33a`
|
| 179 |
(2026-10-30), the eval screen at `e6f331c` and the write at `a400dfe` (2026-11-01).
|
| 180 |
The clustering stage's `summary.json` entry was written by hand from that run's record.
|
| 181 |
+
The deepclust70 eval/test screen ran at `ed1d75b` (2026-11-02) and the return of unused
|
| 182 |
+
held-out reps at `fa6d689` (2026-11-04); MMseqs2 18.8cc5c throughout.
|
| 183 |
|
| 184 |
## License and attribution
|
| 185 |
|
deepclust70/returned.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"params": {
|
| 3 |
+
"src": "/home/bryanbriney/data/tmp/deepclust70-heldout/orig",
|
| 4 |
+
"uniref70": "/home/bryanbriney/data/oplm-sequence-data-dev/uniref70",
|
| 5 |
+
"seed": 0,
|
| 6 |
+
"screen_args": [
|
| 7 |
+
"--min-seq-id",
|
| 8 |
+
"0.7",
|
| 9 |
+
"-c",
|
| 10 |
+
"0.8",
|
| 11 |
+
"--cov-mode",
|
| 12 |
+
"0",
|
| 13 |
+
"-s",
|
| 14 |
+
"1",
|
| 15 |
+
"--max-seqs",
|
| 16 |
+
"100000",
|
| 17 |
+
"-e",
|
| 18 |
+
"1e6"
|
| 19 |
+
]
|
| 20 |
+
},
|
| 21 |
+
"stats": {
|
| 22 |
+
"hits_deepclust70": 0,
|
| 23 |
+
"hits_uniref70": 5,
|
| 24 |
+
"singletons": {
|
| 25 |
+
"unused": 155000,
|
| 26 |
+
"returned": 154998
|
| 27 |
+
},
|
| 28 |
+
"gte2": {
|
| 29 |
+
"unused": 155000,
|
| 30 |
+
"returned": 154997
|
| 31 |
+
},
|
| 32 |
+
"shard": {
|
| 33 |
+
"name": "shard-05353.parquet",
|
| 34 |
+
"rows_before": 586520,
|
| 35 |
+
"rows_after": 896515
|
| 36 |
+
}
|
| 37 |
+
},
|
| 38 |
+
"mmseqs_version": "18.8cc5c",
|
| 39 |
+
"oplm_version": "0.4.1.dev268+gfa6d68911"
|
| 40 |
+
}
|
deepclust70/test_extra_gte2.parquet
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:7a5b43d73f1f546a9ed09872adaad1e326ea9a3d5214bfe93bf05877a73f8b28
|
| 3 |
-
size 8928524
|
|
|
|
|
|
|
|
|
|
|
|
deepclust70/test_extra_singletons.parquet
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:12b3e31103c046c03fb3de2992702df8aa986f6726a5e848f4c3e23edb3740fe
|
| 3 |
-
size 7113390
|
|
|
|
|
|
|
|
|
|
|
|
deepclust70/train/shard-05353.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0b7c081b66e02f9fb3e1eb35789279d783b1aba2479a24fe020b69fa8deec03e
|
| 3 |
+
size 81311284
|
manifest.parquet
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08bcbff57dfc849a2963e75a07da38291db488aeb3778de65f33b3b411c0d35f
|
| 3 |
+
size 496617
|