search

YoshitakaMo / localcolabfold

ColabFold on your local PC
MIT License
607 stars 134 forks source link

Question:pLDDT=nan pTM=nan ipTM=nan error with large complexes #164

Open fglaser opened 1 year ago

fglaser commented 1 year ago

What is your question? I get pLDDT=nan pTM=nan ipTM=nan with AF-multimer only for large complexes, above 1200 aa more or less With smaller complexes it runs perfectly OK.

I updated the colabfold version and tried different version of multimer, nothing helps....

Could you please suggest a way to solve it? All the requested information below.

Tahnks! Fabian

Computational environment Ubuntu 20.04.6 LTS (GNU/Linux 5.15.0-56-generic x86_64)

Colabfold version (updated today) colabfold 1.5.2 (f59147dbfb7b4be95178da1aa6b18b839a2ebee0)

To Reproduce

input sh file

!/bin/bash

export CUDA_VISIBLE_DEVICES=2

colabfold_batch \ --model-type alphafold2_multimer_v3 \ --num-recycle 10 \ --rank multimer \ --save-recycles \ --overwrite-existing-results \ ADR1-ADR2.fasta \ ADR1-ADR2

fasta file more ADR1-ADR2.fasta

ADR1-ADR2 MDQNPNYNFGYGQAYGSGTDHTSDSTNNYNWASQWSQPESAASLATHTFPQYVSQQQQQQQQQAQQQAQNTYAAMNPISTFMQQQQRAQTFPQKKYGQQGGAPKPSAIRNNNFGAFGGGHALSQEWVQPMSQNQMGG PQGNRFFNNQKGGPFNQNKPNWRQNKPKGPAAPKKFDSTGKSPAMLLHELFKDVSEEYTEVEGVPKKYCCTLKVNGRTFQMESVNKKAAKQKCSELVVRDLRPDVHVTPFEEGVAAKAAAPVKKEIDAASGNGQNNK RNLLQADAISNQPTPKKVSAVKKAKLQLTPVESALSLLDLMQKIIAESAEKYSPVFEASEVPKDPEIPEVEVKKEEVDTNGENVANEKKSGWRKNETMHNVTLKFVEQNKQYTKMGPSRGVLKDMVIREALRDLFNV SHADITTVARRHASNRLGHDTTILQCLNTICSILNCTLTIECEPAEDRPLGIGRAYFMAKCTIIDHNENDLKFEVKSSSLASKAMAKDWVAQETLKNYFAIDPSSCVKTDAVSSQGPCALLHAMLNKQTKQKCKIAY EFKDNVPPVAGQATTTFYCECVIDETDRYIGVGRSKKLAKSEAAMQALKKLFKIDYDPAGNYPLALTSRAMTESKVSPLCRHIAEFCKREYHQMTEYYQIPPSNLFAAFLLVNAQEEKRVLAMGSSIQYIVEPDTLS GANGTSLLHLDAIILARRAMLKAFIHELSTVDSECSIFEKKEEGKAALKPNLRLVLYSNYSPPCIHAVDDAATKKLSYVTPTNLTCVPDDVLTYEQIKETKSLRVHCTADKLFKWNTLGIQGALLSNVLHPIFIDNI FFGSEAPVSDESLSYALQGRLGPNENEREIIVESMPVQMRMHMGISHLWHRGVDSVETLDYNTGRTSKGSPSRVCKAEIFEAYRKLNGVDQAVVNYAKAKEMASEYQYEKKVFYEKLEAAGLGKWQTKPAELVDSFT LAAFD:MSVEEGMEVELKSEKMDLDDNIPDFVKETVERSGKNPMSLFSELYVHMTGNTPVFDFYNRNQPNGSMKFICVVILNGERIEGNVKSKKKEAKVSCSLKGLEVVLKHVSEYVVPAVKFEEKTTFFEMLREHT YAKFYELCKNNALIYGFEKVIASVFLKINGNLQIIALSTGNKGLRGDKIVNDGTALIDCHAEILARRGLLRFLYSEVLKFSTEPPNSIFTKGKNALVLKPGISFHLFINTAPCGVARIDKKLKPGTSDDLQNSSRLR FKIDKGMGTVLGGASEFEAPQTFDGIMMGERMRTMSCSDKLLRANVLGVQGAILSHFIDPIYYSSIAVAELNNADRLRKAVYSRAATFKPPAPFHVQDVEIGECQVEDTEQSTSAAARSTISSMNWNLADGNTEVVR TSDGMVHDKDMSGADITTPSRLCKKNMAELMITICTLTKTSVDYPISYEELKAGSQEYAAAKKSFITWLRQKDLGIWQRKPREFQMFTIN

log.txt file: 2023-07-11 10:45:53,727 Running colabfold 1.5.2 (f59147dbfb7b4be95178da1aa6b18b839a2ebee0) 2023-07-11 10:49:58,495 Running on GPU 2023-07-11 10:49:58,891 Found 4 citations for tools or databases 2023-07-11 10:49:58,891 Query 1/1: ADR1-ADR2 (length 1459) 2023-07-11 10:50:07,780 Setting max_seq=252, max_extra_seq=1152 2023-07-11 10:54:21,474 alphafold2_multimer_v2_model_1_seed_000 recycle=0 pLDDT=nan pTM=nan ipTM=nan 2023-07-11 10:57:59,000 alphafold2_multimer_v2_model_1_seed_000 recycle=1 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:01:35,626 alphafold2_multimer_v2_model_1_seed_000 recycle=2 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:05:11,966 alphafold2_multimer_v2_model_1_seed_000 recycle=3 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:08:48,062 alphafold2_multimer_v2_model_1_seed_000 recycle=4 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:12:24,336 alphafold2_multimer_v2_model_1_seed_000 recycle=5 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:16:00,328 alphafold2_multimer_v2_model_1_seed_000 recycle=6 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:19:36,973 alphafold2_multimer_v2_model_1_seed_000 recycle=7 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:23:13,249 alphafold2_multimer_v2_model_1_seed_000 recycle=8 pLDDT=nan pTM=nan ipTM=nan tol=nan 2023-07-11 11:26:49,475 alphafold2_multimer_v2_model_1_seed_000 recycle=9 pLDDT=nan pTM=nan ipTM=nan tol=nan

Expected behavior A clear and concise description of what you expected to happen.

fglaser commented 1 year ago

Hi,

I think it is related to the size of the multimer, over about 1500 aa it does not work, below that works perfectly OK. Any suggestion how to solve it? Thanks! Fabian Technion, Israel

YoshitakaMo commented 1 year ago

When I ran the prediction for the sequence you showed in my Ubuntu 22.04, it didn't show nan values. Here is the log:

$ colabfold_batch \
  --amber \
  --template \
  --use-gpu-relax \
  --num-recycle 3 \
  --overwrite-existing-results \
  --random-seed 0 \
  ADR1-ADR2.fasta \
  ADR1_2

2023-07-16 15:54:17,197 Running colabfold 1.5.2 (f59147dbfb7b4be95178da1aa6b18b839a2ebee0)
2023-07-16 15:54:17.816696: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
2023-07-16 15:54:18,164 Running on GPU
2023-07-16 15:54:18,564 Found 8 citations for tools or databases
2023-07-16 15:54:18,564 Query 1/1: ADR1-ADR2 (length 1459)
COMPLETE: 100%|███| 300/300 [elapsed: 00:02 remaining: 00:00]
2023-07-16 15:54:40,657 Sequence 0 found templates: ['8e4x_B', '6vff_B', '5hp2_D', '1zy7_A', '6d06_A', '8e0f_A', '7kfn_D', '1zy7_B', '6d06_D', '7zpk_C', '5dv7_C', '5n8l_A', '2l3j_A', '2b7v_A', '2l2k_B', '2l3j_A', '2l33_A', '3p1x_A', '2b7v_A', '2l3j_A']
2023-07-16 15:54:44,634 Sequence 1 found templates: ['8e4x_B', '6vff_B', '5hp2_D', '6d06_A', '8e0f_A', '1zy7_A', '1zy7_B', '7kfn_D', '6d06_D', '2l3j_A', '2l33_A', '2l2k_B', '2b7v_A', '2l3j_A', '2mdr_A', '3p1x_B', '3p1x_A', '5dv7_C', '2b7t_A', '2l3c_A']
COMPLETE: 100%|███| 300/300 [elapsed: 00:02 remaining: 00:00]
2023-07-16 15:54:48,024 Setting max_seq=508, max_extra_seq=2048
2023-07-16 15:56:50,264 alphafold2_multimer_v3_model_1_seed_000 recycle=0 pLDDT=63.7 pTM=0.41 ipTM=0.201
2023-07-16 15:58:32,148 alphafold2_multimer_v3_model_1_seed_000 recycle=1 pLDDT=63.4 pTM=0.381 ipTM=0.178 tol=20.2
2023-07-16 16:00:14,503 alphafold2_multimer_v3_model_1_seed_000 recycle=2 pLDDT=63.5 pTM=0.373 ipTM=0.196 tol=8.33
2023-07-16 16:01:57,114 alphafold2_multimer_v3_model_1_seed_000 recycle=3 pLDDT=64.1 pTM=0.371 ipTM=0.188 tol=4.55
2023-07-16 16:01:57,115 alphafold2_multimer_v3_model_1_seed_000 took 426.4s (3 recycles)
2023-07-16 16:03:40,892 alphafold2_multimer_v3_model_2_seed_000 recycle=0 pLDDT=64.4 pTM=0.383 ipTM=0.19
2023-07-16 16:05:23,607 alphafold2_multimer_v3_model_2_seed_000 recycle=1 pLDDT=64.9 pTM=0.386 ipTM=0.208 tol=8.31
2023-07-16 16:07:06,350 alphafold2_multimer_v3_model_2_seed_000 recycle=2 pLDDT=64.4 pTM=0.371 ipTM=0.203 tol=10.7

I used templates and alphafold2_multimer_v3 model for the prediction, but you used the alphafold2-multimer-v2 model alphafold2_multimer_v2_model_1. What happens when you change these settings?

fglaser commented 1 year ago

Hi, Thanks, I tried both, same result... it was just one of the test I did.

Weird, any suggestion? Thansk! Fabian

fglaser commented 1 year ago

Thanks again,

I tried in a colleague server similar to min and it works.

Any suggestion where to look into?

Thanks a lot!

Regards, Fabian

Fabian Glaser, PhD

Structural and Computational Biology Unit The Lorry I. Lokey Center for Life Sciences and Engineering Technion - Israel Institute of Technology, Haifa, Israel

On 16 Jul 2023, at 10:13, Yoshitaka Moriwaki @.***> wrote:

When I ran the prediction for the sequence you showed in my Ubuntu 22.04, it didn't show nan values. Here is the log:

$ colabfold_batch \ --amber \ --template \ --use-gpu-relax \ --num-recycle 3 \ --overwrite-existing-results \ --random-seed 0 \ ADR1-ADR2.fasta \ ADR1_2

2023-07-16 15:54:17,197 Running colabfold 1.5.2 (f59147dbfb7b4be95178da1aa6b18b839a2ebee0) 2023-07-16 15:54:17.816696: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT 2023-07-16 15:54:18,164 Running on GPU 2023-07-16 15:54:18,564 Found 8 citations for tools or databases 2023-07-16 15:54:18,564 Query 1/1: ADR1-ADR2 (length 1459) COMPLETE: 100%|███| 300/300 [elapsed: 00:02 remaining: 00:00] 2023-07-16 15:54:40,657 Sequence 0 found templates: ['8e4x_B', '6vff_B', '5hp2_D', '1zy7_A', '6d06_A', '8e0f_A', '7kfn_D', '1zy7_B', '6d06_D', '7zpk_C', '5dv7_C', '5n8l_A', '2l3j_A', '2b7v_A', '2l2k_B', '2l3j_A', '2l33_A', '3p1x_A', '2b7v_A', '2l3j_A'] 2023-07-16 15:54:44,634 Sequence 1 found templates: ['8e4x_B', '6vff_B', '5hp2_D', '6d06_A', '8e0f_A', '1zy7_A', '1zy7_B', '7kfn_D', '6d06_D', '2l3j_A', '2l33_A', '2l2k_B', '2b7v_A', '2l3j_A', '2mdr_A', '3p1x_B', '3p1x_A', '5dv7_C', '2b7t_A', '2l3c_A'] COMPLETE: 100%|███| 300/300 [elapsed: 00:02 remaining: 00:00] 2023-07-16 15:54:48,024 Setting max_seq=508, max_extra_seq=2048 2023-07-16 15:56:50,264 alphafold2_multimer_v3_model_1_seed_000 recycle=0 pLDDT=63.7 pTM=0.41 ipTM=0.201 2023-07-16 15:58:32,148 alphafold2_multimer_v3_model_1_seed_000 recycle=1 pLDDT=63.4 pTM=0.381 ipTM=0.178 tol=20.2 2023-07-16 16:00:14,503 alphafold2_multimer_v3_model_1_seed_000 recycle=2 pLDDT=63.5 pTM=0.373 ipTM=0.196 tol=8.33 2023-07-16 16:01:57,114 alphafold2_multimer_v3_model_1_seed_000 recycle=3 pLDDT=64.1 pTM=0.371 ipTM=0.188 tol=4.55 2023-07-16 16:01:57,115 alphafold2_multimer_v3_model_1_seed_000 took 426.4s (3 recycles) 2023-07-16 16:03:40,892 alphafold2_multimer_v3_model_2_seed_000 recycle=0 pLDDT=64.4 pTM=0.383 ipTM=0.19 2023-07-16 16:05:23,607 alphafold2_multimer_v3_model_2_seed_000 recycle=1 pLDDT=64.9 pTM=0.386 ipTM=0.208 tol=8.31 2023-07-16 16:07:06,350 alphafold2_multimer_v3_model_2_seed_000 recycle=2 pLDDT=64.4 pTM=0.371 ipTM=0.203 tol=10.7 I used templates and alphafold2_multimer_v3 model for the prediction, but you used the alphafold2-multimer-v2 model alphafold2_multimer_v2_model_1. What happens when you change these settings?

— Reply to this email directly, view it on GitHub https://github.com/YoshitakaMo/localcolabfold/issues/164#issuecomment-1637002522, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACSBVSH4WSEIMZ6UVD4NMC3XQOICNANCNFSM6AAAAAA2FYR7OY. You are receiving this because you authored the thread.

YoshitakaMo commented 1 year ago

Have you tried the reinstallation after deleting the current localcolabfold? If it was installed some time ago, it might be worth a try.

fglaser commented 1 year ago

Hi,

Yes I reinstalled, but nothing changed.

Maybe some memory issue?

Regards, Fabian

Fabian Glaser, PhD

Structural and Computational Biology Unit The Lorry I. Lokey Center for Life Sciences and Engineering Technion - Israel Institute of Technology, Haifa, Israel

On 16 Jul 2023, at 19:17, Yoshitaka Moriwaki @.***> wrote:

Have you tried the reinstallation after deleting the current localcolabfold? If it was installed some time ago, it might be worth a try.

— Reply to this email directly, view it on GitHub https://github.com/YoshitakaMo/localcolabfold/issues/164#issuecomment-1637130643, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACSBVSCJAIUMXNPZ5WNBATTXQQHYLANCNFSM6AAAAAA2FYR7OY. You are receiving this because you authored the thread.