You are here

Where to find old version of Rosetta with git version number

3 posts / 0 new
Last post
Where to find old version of Rosetta with git version number
#1

Hi, there:
    I am reading some papers using Rosetta for modeling work. In the papers, they mentioned rosetta git version number. I am just wondering  if anyone can tell me where I can find some old version of rosetta according to git version number. Thank you very much for your help.

 

Best,

Mao

Category: 
Post Situation: 
Wed, 2019-08-07 01:22
mao_x

On the individual download pages for each version on the RosettaCommons website (accessible from https://www.rosettacommons.org/software/license-and-download) there should be a Git SHA1 hash number which corresponds to the Git version number for that particular release.

That's not necessarily helpful in itself, as there's not really a before/after correspondence between the Git SHA1 numbers. That's where the revision map (https://benchmark.graylab.jhu.edu/tools/revision_map) can come in handy. You can search that page for your particular revision (e.g. 627f7dd22223c3074594934b789abb) and see the date and master number that corresponds to that version. (e.g. 2017-01-31 09:09:55 master:59247). You can then look at a particular released revision and see where that falls, whether it's earlier or later than that revision, and by how much.

For example, from the download page for the 2017 week 8 release (Rosetta 2017.08), you find the SHA1 for that release is 74d9e0b99ea4c80a95cd5b49e2cf71cffe98c85c. From the revision map, you can find that it's later in both date and master number than the version you're interested in (2017-02-24 14:44:24 master:59291). As such it should contain all the code that was in your intended revison. However, if you have Rosetta 3.7, you can find the corresponding SHA1 from the download page (9005cc64587b4189882337bc87783ab96ead263f) and find it's earlier in date and master number than the version in question (2016-08-09 10:33:18 master:58837). As such, it might not contain all the features of the version you're searching for.

Note, though, that not all of the version numbers that are on the revision map page are available as downloads from the RosettaCommons website. (Most of them are not.) In that case, you probably want to look for a weekly release which is slightly after the version specified. Most versions after a given one should have all the features of that particular revision. (And often there's not too substantial change from one weekly release to the next.) Generally speaking, there isn't much issue with using a version of Rosetta which is newer than a particular revision. (Using one which is older often means you're missing features, but a newer version should contain all the necessary code.) The code typically doesn't change too much over a few weeks, so it's unlikely that the slight difference between the versions will be significant in recapitulating a result.

Sometimes, though, you'll find that someone might specify a SHA1 which isn't present on the revision map at all. For those cases, it's usually because the person was using an internal development build of Rosetta, and not one of the "master" builds. In those situations, the best course of action is to post the SHA1 on the forums, and a developer can help you track down which particular released version of Rosetta has the code which that SHA1 corresponds to, or at least the closest available equivalent.

 

As a final note, I'll say that in most cases it often isn't necessary to match the exact version of Rosetta that is being used. If all you want to do is repeat a protocol that has been publishes on a new system, you are often perfectly fine with using the most recent version of Rosetta. Protocols will often work just the same (or better) with the most recent version of Rosetta than with the version used in the published paper. Even if there are issues with running the protocol, it's often just a matter of adjusting a few flags to get all the benefits of the most recent Rosetta. Personally, I'd only worry about trying to match an older version of Rosetta if you've already tried the most recent version and just can't get it to work, or (more critically) keep ending up with results which don't seem to scientifically match the results being published, and wish to dig into that further.

Wed, 2019-08-07 07:27
rmoretti

Thank you for your suggestion.

Thu, 2019-08-08 01:37
mao_x