init

.gitignore

@@ -0,0 +1,11 @@
+.*
+**/__pycache__/
+!.gitignore
+*.out
+logs*/
+results/
+Subtype-GAN/
+data/
+.Rproj.user
+*.Rproj
+Depracated/

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2022 Zhiwei Rong
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

conf/config.yaml

@@ -0,0 +1,37 @@
+# 关于hydra的设置
+hydra:
+  job:
+    # 将工作路径设置到自动生成的路径中
+    chdir: true
+  run:
+    # 设置自动生成路径的位置
+    dir: /mnt/data1/share_data/rongzhiwei/outputs/${now:%Y-%m-%d}/${now:%H-%M-%S}
+
+defaults:
+  - dataset: pan
+  - model: big
+  - train: ${dataset}
+  - _self_
+
+dataset:
+  # 组成batch的方式
+  iter_method: "loader"  # "random"
+  # 数据预处理：标准化的组学
+  normalize: ["meth", "rna", "CN", "miRNA"]
+  # 数据预处理：限制值的范围
+  clip: null
+  # 数据预处理：根据方差进行变量筛选
+  select: null
+  # 组学所服从的分布，会影响重构误差的形式
+  omic_dist_kind: "normal_varconst"
+  # omic_dist_kind:
+  #   meth: "binary_conti"
+  #   rna: "normal_varconst"
+  #   CN: "laplace_varconst"
+  #   miRNA: "normal_varconst"
+
+seed: 1
+cv: null
+test_size: 0.2  # 如果test_size是NULL且cv>1，则test_size默认是0.1
+
+save_model: false

conf/dataset/pan.yaml

@@ -0,0 +1,9 @@
+dat_name: "PAN"
+root: "/mnt/data1/share_data/TCGA/PAN2"
+csvfn:
+  meth: "meth.csv"
+  rna: "rna.csv"
+  CN: "CN.csv"
+  miRNA: "miRNA.csv"
+  cli: "clinic.csv"
+n_clusters: 32

conf/dataset/sub.yaml

@@ -0,0 +1,19 @@
+dat_name: "BRCA"
+root: "/mnt/data1/share_data/TCGA/SubtypeGAN/${dataset.dat_name}"
+csvfn:
+  meth: "meth.csv"
+  rna: "rna.csv"
+  CN: "CN.csv"
+  miRNA: "miRNA.csv"
+  cli: "clinic.csv"
+n_clusters:
+  BRCA: 5
+  BLCA: 5
+  KIRC: 4
+  GBM: 3
+  LUAD: 3
+  PAAD: 2
+  SKCM: 4
+  STAD: 3
+  UCEC: 4
+  UVM: 4

conf/model/big.yaml

@@ -0,0 +1,19 @@
+omic_cls_hiddens: [512, 512]
+omic_cls_latents: 512
+global_hiddens: [512, 512]
+att_method: "gate"
+att_kwargs:
+  use_tanh: false
+  attention_score_method: "sigmoid"
+add_bn_after_att: true
+gumbel_reparam: true
+y_gen: true
+
+omic_enc_hiddens: [512, 512]
+omic_n_latents: 512
+omic_dec_hiddens: [512, 512]
+omic_enc_inject: true
+omic_dec_inject: false
+prior_kind: "uniform"
+bn: true
+dp: null

conf/model/middle.yaml

@@ -0,0 +1,19 @@
+omic_cls_hiddens: [512]
+omic_cls_latents: 512
+global_hiddens: [512]
+att_method: "gate"
+att_kwargs:
+  use_tanh: false
+  attention_score_method: "sigmoid"
+add_bn_after_att: true
+gumbel_reparam: true
+y_gen: true
+
+omic_enc_hiddens: [512]
+omic_n_latents: 512
+omic_dec_hiddens: [512]
+omic_enc_inject: true
+omic_dec_inject: false
+prior_kind: "uniform"
+bn: true
+dp: null

conf/model/small.yaml

@@ -0,0 +1,23 @@
+omic_cls_hiddens:  []
+omic_cls_latents: 100
+global_hiddens: []
+att_method: "gate"
+att_kwargs:
+  use_tanh: false
+  attention_score_method: "sigmoid"
+add_bn_after_att: true
+gumbel_reparam: true
+y_gen: true
+
+omic_enc_hiddens: [250]
+omic_n_latents:
+  rna: 100
+  CN: 100
+  meth: 100
+  miRNA: 30
+omic_dec_hiddens: [100]
+omic_enc_inject: true
+omic_dec_inject: false
+prior_kind: "uniform"  # [0.9, 0.1]
+bn: true
+dp: null

conf/train/pan.yaml

@@ -0,0 +1,21 @@
+n_epoch: 500
+bs: 512
+lr: 0.0005
+nw: 8
+device: cuda:0
+tensorboard: "PAN"
+evaluation: 1
+verbose: null
+early_stop:
+  metric: "loss"
+  patience: 10
+  verbose: false
+  delta: 1.0
+lr_sch: true
+optim_method: "adam"
+weight_decay: 0.
+kly_anneal: "cosine"
+kly_anneal_args: {"maxv": 10, "minv": 3}
+rec_anneal: "constant"
+rec_anneal_args: {"minv": 1.}
+kl_weight: 1.

conf/train/sub.yaml

@@ -0,0 +1,21 @@
+n_epoch: 500
+bs: 32
+lr: 0.0005
+nw: 2
+device: cuda:2
+tensorboard: _datname
+evaluation: 1
+verbose: null
+early_stop:
+  metric: "loss"
+  patience: 10
+  verbose: false
+  delta: 1.0
+lr_sch: true
+optim_method: "adam"
+weight_decay: 0.
+kly_anneal: "cosine"
+kly_anneal_args: {"maxv": 10, "minv": 1}
+rec_anneal: "constant"
+rec_anneal_args: {"minv": 1}
+kl_weight: 1.

environment.sh

@@ -0,0 +1,13 @@
+# first, make sure that there is enough disk space
+# then, create a new conda environment
+# finally, run this script
+set -v
+
+conda install python=3.9 -y
+conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch -y
+conda install numpy pandas matplotlib seaborn tqdm scikit-learn -c conda-forge -y
+conda install lifelines optuna -c conda-forge -y
+conda install flake8 ipdb jupyterlab tensorboard -c conda-forge -y
+
+# pip的版本更高
+pip install hydra-core