I think this script matches AJ's best relax flavor from SRC 2020

[smlewis]

Not going to merge without her explicit permission but I'll put it here for comment (it was discussed in slack yesterday)

[AJVincelli]

Thank you Steven! This would have taken me... a really long time... to write. I will test out this script and see if it performs similarly to the command-line version that I presented at SRC 2020, and report back. Give me... some time.

[AJVincelli]

Note that @danielzaidman recently reported that this script results in a slightly misaligned pose, which might affect RMSD. I will also look into this.

[danielzaidman]

I'm still not sure why (looking into it), but as it is at the moment, the result of this script is definitely different than "java", even after aligning the chains back. Java is much more conservative with the backbone. @smlewis @AJVincelli

[danielzaidman]

Hey @smlewis , @AJVincelli . I added a missing line, and now the output looks the same as "java". I think we can merge now.

[roccomoretti]

I think this is entirely @AJVincelli's call as to whether this script properly represents her best-practices settings. I don't see anything immediately at issue with it, but I don't have the full flag-set for Java with me at the moment, so I can't double check.

[smlewis]

It definitely shouldn't merge until closer to her paper.

[AJVincelli]

Well, the good news is that I'm working on Relax things... The bad news is that I haven't gotten to testing this script yet. I'm probably ~1 month away from testing and comparing with previous results. Even if the manuscript isn't mature and it won't be a good time to merge, I can still report back with a stamp of approval. Is that timeline acceptable?

[AJVincelli]

The main Java flags are -constrain_relax_to_start_coords -relax:ramp_constraints false -relax:script MonomerRelax2019 (I think Daniel's small fix added the MonomerRelax2019 equivalent?). And technically, Java uses -nstruct 5 (and -in:file:fullatom but I don't know if that actually does anything).

[AJVincelli]

In the scripts_to_parse.py file, is the / at the beginning of /scripts/public/preparation/constrained_relax.xml intentional?

[AJVincelli]

Hm, I'm getting an error when I include the commandline flag -out:mmCIF. I've asked Slack how I can specify a CIF file output using RosettaScripts. I'll post back when I have updates...

[AJVincelli]

It appears that the -out:mmCIF flag should work. Can anybody get this script to run with the -out:mmCIF flag?

[AJVincelli]

Rocco helped me debug the issue with -out:mmCIF. If I also add the flag -out::file::output_pose_cache_data false, it works! I will now test this script on my 500-structure dataset, and report back if it works as expected and reproduces my previous results. (Can you tell that I'm trying to get this before RosettaCON??)

[roccomoretti]

The issue with -out:mmCIF is RosettaCommons/main#5503

[AJVincelli]

Hi all, I've (finally!) tested this script with my benchmark dataset, and it works beautifully with Daniel's small fix and the -out::file::output_pose_cache_data false flag. I also removed the / at the beginning of the /scripts/public/preparation/constrained_relax.xml line (see above), I'm not sure if that matters. I'm hoping to write up the manuscript this autumn, so we're getting closer to merging...

[weitzner]

It seems interesting that there is not any cartesian space minimization used here. Was that tested as part of this work?

[AJVincelli]

Hi @weitzner, yes! All of the juicy details are here: http://bit.ly/BestRelaxMethod I was surprised to find that Cartesian Relax didn’t improve the Quality Scores as much (such as the Latte flavor).

[weitzner]

Interesting! Thanks for passing the doc along!

[AJVincelli]

You are very welcome, I’m glad it’s helpful! Let me know if you have any questions/thoughts. AJ

…

________________________________ From: Brian Weitzner ***@***.***> Sent: Thursday, December 2, 2021 10:37 AM To: RosettaCommons/rosetta_scripts_scripts Cc: AJ Vincelli; Mention Subject: Re: [RosettaCommons/rosetta_scripts_scripts] I think this script matches AJ's best relax flavor from SRC 2020 (#14) Interesting! Thanks for passing the doc along! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#14 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ALC2NUI7DBFHD2AO2QFGOGTUO6HBZANCNFSM4QGUEC6Q>. Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.

[ajasja]

@AJVincelli How are things going? Do you have plans to merge this? I guess it has to go into the new repo now...