Merge remote-tracking branch 'upstream/main'

alphafold/model/folding_multimer.py

@@ -789,7 +789,7 @@ def backbone_loss(gt_rigid: geometry.Rigid3Array,
   loss_fn = functools.partial(
       all_atom_multimer.frame_aligned_point_error,
       l1_clamp_distance=config.atom_clamp_distance,
-      loss_unit_distance=config.loss_unit_distance)
+      length_scale=config.loss_unit_distance)
 
   loss_fn = jax.vmap(loss_fn, (0, None, None, 0, None, None, None))
   fape = loss_fn(target_rigid, gt_rigid, gt_frames_mask,

alphafold/model/utils.py

@@ -34,7 +34,7 @@ def final_init(config):
 
 def batched_gather(params, indices, axis=0, batch_dims=0):
   """Implements a JAX equivalent of `tf.gather` with `axis` and `batch_dims`."""
-  take_fn = lambda p, i: jnp.take(p, i, axis=axis)
+  take_fn = lambda p, i: jnp.take(p, i, axis=axis, mode='clip')
   for _ in range(batch_dims):
     take_fn = jax.vmap(take_fn)
   return take_fn(params, indices)

docker/Dockerfile

@@ -18,13 +18,10 @@ FROM nvidia/cuda:${CUDA_FULL}-base-ubuntu20.04 AS build
 # previously set).
 ARG CUDA_FULL
 ARG CUDA=11.2
-# JAXLIB no longer built for all minor CUDA versions:
-# https://github.com/google/jax/blob/main/CHANGELOG.md#jaxlib-0166-may-11-2021
-ARG CUDA_JAXLIB=11.1
 ARG DEBIAN_FRONTEND=noninteractive
 
 # Use bash to support string substitution.
-SHELL ["/bin/bash", "-c"]
+SHELL ["/bin/bash", "-o", "pipefail", "-c"]
 
 RUN apt-get update \
     && apt-get upgrade -y \
@@ -71,11 +68,11 @@ RUN wget -q -P /app/alphafold/alphafold/common/ \
   https://git.scicore.unibas.ch/schwede/openstructure/-/raw/7102c63615b64735c4941278d92b554ec94415f8/modules/mol/alg/src/stereo_chemical_props.txt
 
 # Install pip packages.
-RUN pip3 install --upgrade --no-cache-dir pip \
-    && pip3 install --no-cache-dir -r /app/alphafold/requirements.txt \
+RUN pip3 install --upgrade pip --no-cache-dir \
+    && pip3 install -r /app/alphafold/requirements.txt --no-cache-dir \
     && pip3 install --upgrade --no-cache-dir \
-      jax==0.2.14 \
-      jaxlib==0.1.69+cuda$(cut -f1,2 -d. <<< ${CUDA_JAXLIB} | sed 's/\.//g') \
+      jax==0.3.17 \
+      jaxlib==0.3.15+cuda11.cudnn805 \
       -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
 
 # Apply OpenMM patch.

docker/requirements.txt

@@ -1,3 +1,3 @@
 # Dependencies necessary to execute run_docker.py
-absl-py==0.13.0
+absl-py==1.0.0
 docker==5.0.0

notebooks/AlphaFold.ipynb

@@ -478,6 +478,12 @@
         "\n",
         "run_relax = True  #@param {type:\"boolean\"}\n",
         "\n",
+        "#@markdown Relaxation is faster with a GPU, but we have found it to be less stable.\n",
+        "#@markdown You may wish to enable GPU for higher performance, but if it doesn't\n",
+        "#@markdown converge we suggested reverting to using without GPU.\n",
+        "\n",
+        "relax_use_gpu = False  #@param {type:\"boolean\"}\n",
+        "\n",
         "# --- Run the model ---\n",
         "if model_type_to_use == notebook_utils.ModelType.MONOMER:\n",
         "  model_names = config.MODEL_PRESETS['monomer'] + ('model_2_ptm',)\n",
@@ -554,7 +560,7 @@
         "        stiffness=10.0,\n",
         "        exclude_residues=[],\n",
         "        max_outer_iterations=3,\n",
-        "        use_gpu=True)\n",
+        "        use_gpu=relax_use_gpu)\n",
         "    relaxed_pdb, _, _ = amber_relaxer.process(prot=unrelaxed_proteins[best_model_name])\n",
         "  else:\n",
         "    print('Warning: Running without the relaxation stage.')\n",

requirements.txt

@@ -1,14 +1,14 @@
-absl-py==0.13.0
+absl-py==1.0.0
 biopython==1.79
 chex==0.0.7
-dm-haiku==0.0.4
+dm-haiku==0.0.7
 dm-tree==0.1.6
 docker==5.0.0
 immutabledict==2.0.0
-jax==0.2.14
+jax==0.3.17
 ml-collections==0.1.0
-numpy==1.19.5
+numpy==1.21.6
 pandas==1.3.4
 protobuf==3.20.1
 scipy==1.7.0
-tensorflow-cpu==2.5.0
+tensorflow-cpu==2.9.0

setup.py

@@ -18,7 +18,7 @@
 
 setup(
     name='alphafold',
-    version='2.2.3',
+    version='2.2.4',
     description='An implementation of the inference pipeline of AlphaFold v2.0.'
     'This is a completely new model that was entered as AlphaFold2 in CASP14 '
     'and published in Nature.',