models/Divergent421/model.yaml at master · kipoi/models · 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
defined_as: kipoi.model.KerasModel
args:
    weights:
        url: https://zenodo.org/record/1452399/files/model.weights.h5?download=1
        md5: 2a0ae0a29337eb8106d65e1baeda85d1
    arch:
        url: https://zenodo.org/record/1452399/files/model.arch?download=1
        md5: 6903bcab337a6753ad010f43f208df42
    backend: tensorflow
    image_dim_ordering: tf

info:
    authors:
        - name: Nancy Xu
          github: xnancy
          email: xnancy@stanford.edu
    trained_on: "Chromosomes 1, 8, and 21 are test set, 9 is validation set, the remaining data is training data."
    doc: >
      Large-scale multi-task convolutional model for predicting chromatin accessility model.

      Model was trained genome-wide accessibility measures across 421 biosamples (cell lines or tissues) from Roadmap and ENCODE.
    tags:
        - DNA accessibility

default_dataloader:
    defined_as: kipoiseq.dataloaders.SeqIntervalDl
    default_args:
        auto_resize_len: 1000

dependencies:
    conda:
        - h5py=2.10.0
        - tensorflow<=1.4.1
        - keras=1.2.2
        - bioconda::pysam=0.15.3
        - bioconda::kipoiseq
        - python=3.6.15
        - pip=21.3
schema:
    inputs:
        shape: (1000,4)
        doc: "1000 base pair sequence of one-hot encoding ACGT"
    targets:
        shape: (421,)
        doc: "Binary 0/1 output for chromatin accessibility in the designated range. 0 = inaccessible, 1 = accessible."
        column_labels: task_names.txt
test:
  expect:
    url: https://s3.eu-central-1.amazonaws.com/kipoi-models/predictions/14f9bf4b49e21c7b31e8f6d6b9fc69ed88e25f43/Divergent421/predictions.h5
    md5: 62da0ac731f323ea54ee6e30c38e0722
  precision_decimal: 5