"ColabFold v1.5.2-patch: AlphaFold2 using MMseqs2" vs "ColabFold v1.5.2: AlphaFold2 using MMseqs2"

[MarianVincenzi87]

Previously I used the version without patch (i.e., "ColabFold v1.5.2: AlphaFold2 using MMseqs2") to run the modeling pf the structure of my proteins. The problem is that when I run the modeling starting from the same sequence but using the version "ColabFold v1.5.2-patch: AlphaFold2 using MMseqs2" I obtain different result and thus I am not able to reproduce the data acquired previously.

In addition, even using the same version "ColabFold v1.5.2-patch: AlphaFold2 using MMseqs2" for two runs with the same sequence I obtain two different results.

My questions are:

1. Is there anyway to use again "ColabFold v1.5.2: AlphaFold2 using MMseqs2" version?

2. I there a specific parameter to make reproducible two runs starting from the same sequence?

Sorry to bother you and thank you in advance.

Best regards,

Marian Vincenzi

[sokrypton]

Can you provide more details? Do you have the zip output files from both runs? Can you share these with me? [email protected]

[MarianVincenzi87]

Atteched the following files:

EphA2_ColabFold v1.5.2-patch AlphaFold2 using MMseqs2.rar

EphA2_ColabFold v1.5.2 AlphaFold2 using MMseqs2.rar

these are the output realated to the same sequence and same parameters but with the two different versions (ColabFold v1.5.2 and ColabFold v1.5.2-patch).

Thank you again,

Best regards,

Marian Vincenzi

Can you provide more details? Do you have the zip output files from both runs? Can you share these with me? @.***

-- Reply to this email directly, view it on GitHub [1], or unsubscribe [2]. You are receiving this because you authored the thread.Message ID: @.***>

Links:

[1] https://github.com/sokrypton/ColabFold/issues/460#issuecomment-1584509279 [2] https://github.com/notifications/unsubscribe-auth/BAGKZO7ROPLI76TRFHLQAJLXKMKMHANCNFSM6AAAAAAZAKLTRI

[MarianVincenzi87]

sorry for the previous email without files.

Atteched the following files:

EphA2_ColabFold v1.5.2-patch AlphaFold2 using MMseqs2.rar

EphA2_ColabFold v1.5.2 AlphaFold2 using MMseqs2.rar

these are the output realated to the same sequence and same parameters but with the two different versions (ColabFold v1.5.2 and ColabFold v1.5.2-patch).

Thank you again,

Best regards,

Marian Vincenzi

Can you provide more details? Do you have the zip output files from both runs? Can you share these with me? @.***

-- Reply to this email directly, view it on GitHub [1], or unsubscribe [2]. You are receiving this because you authored the thread.Message ID: @.***>

Links:

[1] https://github.com/sokrypton/ColabFold/issues/460#issuecomment-1584509279 [2] https://github.com/notifications/unsubscribe-auth/BAGKZO7ROPLI76TRFHLQAJLXKMKMHANCNFSM6AAAAAAZAKLTRI

[sokrypton]

Hi Marian, I'm not seeing any big differences...

There are gonna be slight differences. Even changing GPU will cause a slight difference... but the differences should be small.