RosettaFold2 (WIP) Colab crash

[jfbazan]

While trying to run RF2 this morning, I get two separate error messages as the program crashes. The first fault occurs at the early 'Setup RosettaFold' stage (below), and the second at the next "Run RosettaFold' point, apparently while trying to assemble the MSA (error msg at the bottom). Thx in advance for your kind help.

FIRST error msg: install RoseTTAFold2 install hhsuite

OSError Traceback (most recent call last) in

/content/RoseTTAFold2/network/predict.py in 6 from torch.utils import data 7 from parsers import parse_a3m, read_templates, read_template_pdb, parse_pdb ----> 8 from RoseTTAFoldModel import RoseTTAFoldModule 9 import util 10 from collections import namedtuple

8 frames /usr/lib/python3.10/ctypes/init.py in init(self, name, mode, handle, use_errno, use_last_error, winmode) 372 373 if handle is None: --> 374 self._handle = _dlopen(self._name, mode) 375 else: 376 self._handle = handle

OSError: libcudart.so.11.0: cannot open shared object file: No such file or directory

SECOND error msg: jobname: xxxxxxxxxxxxx_X1_43e20 lengths: [167, 116]

NameError Traceback (most recent call last) in <cell line: 60>() 59 os.makedirs(jobname, exist_ok=True) 60 if msa_method == "mmseqs2": ---> 61 get_msa(u_sequences, jobname, mode=pair_mode, max_msa=max_extra_msa) 62 63 elif msa_method == "single_sequence":

NameError: name 'get_msa' is not defined

[milot-mirdita]

~Google upgraded CUDA to version 12 last week and broke a much of things. We fixed everything except the rosettafold notebook as it’s incompatible with CUDA 12. We'll try to get it fixed, but we don’t know when exactly~

Sorry, that ESM is broken, not RF.

[sokrypton]

The notebook should work. Can you try again. It looks like maybe the first cell was not executed?