Adding new method: DRVI (#61)

* script for drvi

* add drvi to depenencies

* add nvida image

* changes after feedback

* working DRVI mehtod

* remove comments

* remove comments, preprocessing

* Update src/methods/drvi/script.py

Co-authored-by: Michaela Müller <[email protected]>

* Update src/methods/drvi/config.vsh.yaml

Co-authored-by: Michaela Müller <[email protected]>

* add changelog entry

---------

Co-authored-by: Michaela Müller <[email protected]>

Author: seohyonkim

CHANGELOG.md

@@ -3,6 +3,7 @@
 ## New functionality
 
 * Added `metrics/kbet_pg` and `metrics/kbet_pg_label` components (PR #52).
+* Added `method/drvi` component (PR #61).
 
 ## Minor changes
 

src/methods/drvi/config.vsh.yaml

@@ -0,0 +1,44 @@
+__merge__: ../../api/comp_method.yaml
+name: drvi
+label: DRVI
+summary: "DrVI is an unsupervised generative model capable of learning non-linear interpretable disentangled latent representations from single-cell count data."
+description: |
+  Disentangled Representation Variational Inference (DRVI) is an unsupervised deep generative model designed for integrating single-cell RNA sequencing (scRNA-seq) data across different batches.
+  It extends the variational autoencoder (VAE) framework by learning a latent representation that captures biological variation while disentangling and correcting for batch effects.
+  DRVI conditions both the encoder and decoder on batch covariates, allowing it to explicitly model and mitigate batch-specific variations during training.
+  By incorporating a KL-divergence regularization term, it balances data reconstruction with latent space structure, resulting in a unified embedding where similar cells cluster together regardless of batch.
+references:
+  doi: 
+    - 10.1101/2024.11.06.622266
+links:
+  documentation: https://drvi.readthedocs.io/latest/index.html
+  repository: https://github.com/theislab/DRVI?tab=readme-ov-file
+info:
+  preferred_normalization: counts
+arguments:
+  - name: --n_hvg
+    type: integer
+    default: 2000
+    description: Number of highly variable genes to use.
+  - name: --n_epochs
+    type: integer
+    default: 100
+    description: Number of epochs
+resources:
+  - type: python_script
+    path: script.py
+  - path: /src/utils/read_anndata_partial.py
+engines:
+  - type: docker
+    image: openproblems/base_pytorch_nvidia:1.0.0 
+    setup:
+      - type: python
+        pypi:
+        - drvi-py==0.1.7
+        - torch==2.3.0
+        - torchvision==0.18.0
+runners:
+  - type: executable
+  - type: nextflow
+    directives:
+      label: [midtime,midmem,lowcpu,gpu]

src/methods/drvi/script.py

@@ -0,0 +1,82 @@
+import anndata as ad
+import scanpy as sc
+import drvi
+from drvi.model import DRVI
+from drvi.utils.misc import hvg_batch
+import pandas as pd
+import numpy as np
+import warnings
+import sys
+import scipy.sparse
+
+## VIASH START
+par = {
+  'input': 'resources_test/task_batch_integration/cxg_immune_cell_atlas/dataset.h5ad',
+  'output': 'output.h5ad',
+  'n_hvg': 2000,
+  'n_epochs': 400
+}
+meta = {
+  'name': 'drvi'
+}
+## VIASH END
+
+sys.path.append(meta["resources_dir"])
+from read_anndata_partial import read_anndata
+
+print('Reading input files', flush=True)
+adata = read_anndata(
+    par['input'],
+    X='layers/counts',
+    obs='obs',
+    var='var',
+    uns='uns'
+)
+
+if par["n_hvg"]: 
+     print(f"Select top {par['n_hvg']} high variable genes", flush=True) 
+     idx = adata.var["hvg_score"].to_numpy().argsort()[::-1][:par["n_hvg"]] 
+     adata = adata[:, idx].copy() 
+
+print('Train model with DRVI', flush=True)
+
+DRVI.setup_anndata(
+    adata,
+    categorical_covariate_keys=["batch"],
+    is_count_data=False,
+)
+
+model = DRVI(
+    adata,
+    categorical_covariates=["batch"],
+    n_latent=128,
+    encoder_dims=[128, 128],
+    decoder_dims=[128, 128],
+)
+model
+
+model.train(
+    max_epochs=par["n_epochs"],
+    early_stopping=False,
+    early_stopping_patience=20,
+    plan_kwargs={
+        "n_epochs_kl_warmup": par["n_epochs"],
+    },
+)
+
+print("Store outputs", flush=True)
+output = ad.AnnData(
+    obs=adata.obs.copy(),
+    var=adata.var.copy(),
+    obsm={
+        "X_emb": model.get_latent_representation(),
+    },
+    uns={
+        "dataset_id": adata.uns.get("dataset_id", "unknown"),
+        "normalization_id": adata.uns.get("normalization_id", "unknown"),
+        "method_id": meta["name"],
+    },
+)
+
+print("Write output AnnData to file", flush=True)
+output.write_h5ad(par['output'], compression='gzip')