Do multiple GTX1080 gpu meet the >12GB requirement? and Curious about error at end of first autoAlign run

[btn1234]

I am excited to try emClarity to process tilt series of an in vitro synthesized fiber. I just installed emClarity and have a couple of questions.

1) I have three GTX 1080 gpus. They are not >12 GB alone, but are when taken together. Does that meet the programs requirements?

2) I am starting to work through the tutorial. I created a tilt1.rawtlt file from the information in the tutorial about the data, and upon modifying the example param0.m file using information in the tutorial I attempted to run autoAlign. During the run there was a comment about tilt1.rawtlt being the same file as tilt1.rawtlt, and there is now a tilt1.rawtlt that seems to replace the old one, although the new one can't be opened as a txt file. And at the end of the run there were 2 'No such file or directory' remarks from the program. I'm curious to know if any of these flag problems to which I need to attend, or if they can be ignored?

Attached are the command and screen print that resulted, the param0.m file as a txt file, and the emClarity script file I made also as a txt file.

Thanks in advance for any help to clarify for me.

Tracy Nixon firstRunNotes.txt param0.m.txt emClarity_1_5_2_0_v20b.txt

[bHimes]

Hi Tracy,

memory

• the memory is def okay for 2080 Ti (11.x Gb)
• I think 8 Gb is actually okay, though it will limit the performance. (you'll probably have to restrict nCpuCores to match nGPUs.
  • I can't promis this will work though. The memory on each card is fully independent. Even for cards that do have nvlink, emClarity does not take advantage.

Errors in finding scripts

• From your attached log file, it looks like a dependency was not included properly when 1.5.2.0 was compiled. I'm going to re-label this issue as a bug. I'll try to find some time to sort it out. You can always to a manual tilt-series alignment in IMOD in the mean time. /bin/bash: /dev/shm/emC_tmp_1806211118/.mcrCache9.9/emClar0/mnt/6tb4/apps/emclarity/emClarity/alignment/emC_autoAlign: No such file or directory /bin/bash: /dev/shm/emC_tmp_1806211118/.mcrCache9.9/emClar0/mnt/6tb4/apps/emclarity/emClarity/alignment/emC_findBeads: No such file or directory

[btn1234]

Hi Benjamin,

Thank you for your quick response. We know how to do alignment in etomo, and were doing it for comparison. We've used imod to do subtogram averaging on chunks of these fibers. It's our first tomogram. The calcFSC says 9 Angstroms, but I'm not sure I believe it. We are a molecular biology lab with beginner's experience in cryoEM. Looking to try your program to push things if possible.

Best regards - and thanks for your wonderful sharing of emClarity!

Tracy

---

From: bHimes notifications@github.com Sent: Tuesday, March 9, 2021 9:32 AM To: bHimes/emClarity emClarity@noreply.github.com Cc: Nixon, B Tracy btn1@psu.edu; Author author@noreply.github.com Subject: Re: [bHimes/emClarity] Do multiple GTX1080 gpu meet the >12GB requirement? and Curious about error at end of first autoAlign run (#134)

Hi Tracy,

memory

• the memory is def okay for 2080 Ti (11.x Gb)
• I think 8 Gb is actually okay, though it will limit the performance. (you'll probably have to restrict nCpuCores to match nGPUs.
  • I can't promis this will work though. The memory on each card is fully independent. Even for cards that do have nvlink, emClarity does not take advantage.

Errors in finding scripts

• From your attached log file, it looks like a dependency was not included properly when 1.5.2.0 was compiled. I'm going to re-label this issue as a bug. I'll try to find some time to sort it out. You can always to a manual tilt-series alignment in IMOD in the mean time. /bin/bash: /dev/shm/emC_tmp_1806211118/.mcrCache9.9/emClar0/mnt/6tb4/apps/emclarity/emClarity/alignment/emC_autoAlign: No such file or directory /bin/bash: /dev/shm/emC_tmp_1806211118/.mcrCache9.9/emClar0/mnt/6tb4/apps/emclarity/emClarity/alignment/emC_findBeads: No such file or directory

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FbHimes%2FemClarity%2Fissues%2F134%23issuecomment-793974562&data=04%7C01%7Cbtn1%40psu.edu%7Cead49902b1a74c6feabb08d8e3082ea4%7C7cf48d453ddb4389a9c1c115526eb52e%7C0%7C0%7C637508971572269706%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=kuJgX6aWIpScha%2FBEhdlbrXacPaqjDuI5d6sGWbULRA%3D&reserved=0, or unsubscribehttps://nam10.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAEGUWCXKDGKD2BSHTFUFC4TTCYPPDANCNFSM4Y2UZW6A&data=04%7C01%7Cbtn1%40psu.edu%7Cead49902b1a74c6feabb08d8e3082ea4%7C7cf48d453ddb4389a9c1c115526eb52e%7C0%7C0%7C637508971572279653%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=Vq9VFFmtdJ3j%2Bizc5t3h9Rmvcdo8P%2F6IvsISfAJgVFU%3D&reserved=0.

[bHimes]

A few points to be aware of with filaments:

If you are averaging subtomograms from a fiber that has close repeats, it is pretty easy to inadvertently have density from neighboring subtomos included in each half-set, such that the FSC is invalid because the data is not independent between half-sets.

When you do your "sub-region" selection, you can split a single fiber into multiple tomograms. Then, in your parameter file, set fscGoldSplitOnTomos=1. This will make sure the half-sets are defined on the tomogram level, so the should remain independent.

You can optionally restrict the xyz shifts using mPeak_Radius, which has some description in the example parameter file. This way, neighboring repeats can be used for alignment, but the picked subunits should "stay in place"

[bHimes]

fixed in 1.5.3