limit in protein length or msa sequence number?
Closed this issue · 3 comments
Yogesh1-11 commented
hi as a test case i was using a protein of 539 residue lenth and msa of 25 protein sequences.
but i am getting memory exhaustion error. Is colab limited to protein length or msa size?
patrickbryant1 commented
Hi, the limit is on protein length. A statistical representation is created from the MSA and the size of this thereby doesn't matter. However, 25 sequences is probably too shallow to get good results.
Yogesh1-11 commented
even after chopping protein length to 490 error persists. what is the
maximum protein length for umol? Also, I am using ion(
[O-]P(=O)([O-])[O-]) as ligand.
…---------------------------------------------------------------------------
XlaRuntimeError Traceback (most recent call last)
<ipython-input-6-55efa3eb56f8> <https://localhost:8080/#> in <cell
line: 10>() 8 9 #Predict
---> 10 predict(config.CONFIG, 11 MSA_FEATS, 12
LIGAND_FEATS,
------------------------------
9 frames
------------------------------
[... skipping hidden 14 frame]
/usr/local/lib/python3.10/dist-packages/jax/_src/dispatch.py
<https://localhost:8080/#> in backend_compile(backend, built_c,
options, host_callbacks) 1010 # TODO(sharadmv): remove this
fallback when all backends allow `compile` 1011 # to take in
`host_callbacks`
-> 1012 return backend.compile(built_c, compile_options=options)
1013 1014 # TODO(phawkins): update users.
XlaRuntimeError: RESOURCE_EXHAUSTED: Out of memory while trying to
allocate 8274444288 bytes.
On Sun, Nov 19, 2023 at 2:24 PM Patrick Bryant ***@***.***> wrote:
Closed #5 <#5> as completed.
—
Reply to this email directly, view it on GitHub
<#5 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AZNDTOSAEEZMG6D6N7376GLYFHCKXAVCNFSM6AAAAAA7RRXWBSVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJRGAYDGMZTGMYDCMQ>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
patrickbryant1 commented
This depends on the available RAM. Umol has no limit. I don't know what you have available in Colab, but recommend to run it locally for bigger proteins. We have not trained on phosphorylation sites.