DeepEST/NOTICE at main · BorgwardtLab/DeepEST · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
DeepEST
Copyright (c) the DeepEST authors. See LICENSE.md.

This product includes third-party software and third-party derived data. The files below are not
DeepEST code and are not covered by DeepEST's licence.

DeepEST is itself BSD-3-Clause, so the conditions and disclaimer that the two BSD-3-Clause notices
below require are word-for-word the ones already reproduced in LICENSE.md.

--------------------------------------------------------------------------------
src/deepest/lr_scheduler.py
--------------------------------------------------------------------------------

Contains LinearWarmupCosineAnnealingLR, copied verbatim on 2026-08-24 from:

    Package:  lightning-flash 0.8.1.post0
    File:     flash/core/optimizers/lr_scheduler.py
    Project:  https://github.com/Lightning-AI/lightning-flash
    Original: https://github.com/Lightning-AI/lightning-bolts/blob/master/pl_bolts/optimizers/lr_scheduler.py
    Author:   @ananyahjha93 and the PyTorch Lightning team
    Licence:  Apache License 2.0

    Copyright The PyTorch Lightning team.

    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
    You may obtain a copy of the License at

        http://www.apache.org/licenses/LICENSE-2.0

    Unless required by applicable law or agreed to in writing, software
    distributed under the License is distributed on an "AS IS" BASIS,
    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    See the License for the specific language governing permissions and
    limitations under the License.

Why it is vendored: DeepEST used exactly one symbol from lightning-flash, and that package no
longer resolves cleanly on current Python versions, causing installation to fail for reasons
unrelated to DeepEST. Copying the single class removes the dependency.

What was changed: nothing in the class body. The upstream `_CORE_TESTING` import and the
doctest-skip guard that depended on it were dropped, and an explanatory comment block was added.
`tests/test_lr_scheduler.py` verifies the produced learning-rate sequences match the upstream
implementation exactly, including its behaviour on a degenerate configuration.


--------------------------------------------------------------------------------
DeepFRI
--------------------------------------------------------------------------------

    Project:  https://github.com/flatironinstitute/DeepFRI
    Paper:    Gligorijevic et al., Structure-based protein function prediction using graph
              convolutional networks, Nature Communications 12, 3168 (2021)
    Licence:  BSD 3-Clause License

    Copyright (c) 2021, Flatiron Institute
    All rights reserved.

DeepEST's structure module is transfer-learned from DeepFRI, so DeepFRI material appears here in
three forms:

    preprocessing/gene_function/scripts/src/utils_deepfri.py
        Adapted from DeepFRI's deepfrier/utils.py (PDB and FASTA loading, one-hot encoding).

    preprocessing/gene_function/scripts/src/structure_model.py
        Reimplements DeepFRI's graph-convolutional output layer so its released weights can be
        loaded and its per-protein embeddings extracted.

    assets/deepfri_terms.pkl and examples/borrelia_b31/structures.pkl
        Data derived from DeepFRI release artifacts: the model's GO-term vocabulary, and the
        embeddings its GCN produces for the example species.

The DeepFRI model weights themselves are not redistributed here.
scripts/fetch_reference_data.sh downloads them from the DeepFRI release.


--------------------------------------------------------------------------------
DeepGOPlus
--------------------------------------------------------------------------------

    Project:  https://github.com/bio-ontology-research-group/deepgoplus
    Paper:    Kulmanov and Hoehndorf, DeepGOPlus: improved protein function prediction from
              sequence, Bioinformatics 36(2), 422-429 (2020)
    Licence:  BSD 3-Clause License

    Copyright (c) 2019, Bio-Ontology Research Group
    All rights reserved.

The GO-ontology parsing and hierarchy propagation used to score predictions is adapted from
DeepGOPlus, and appears as:

    src/deepest/utils_deepgoplus.py
    preprocessing/gene_function/scripts/src/utils_deepgoplus.py
    train_model/utils_deepgoplus.py                    (a shim re-exporting the package module)