How do you download old RefSeq releases

[shawn.mek]

If I need to run analysis using a specific release version of Refseq, is there a place to download it (say release 56)?

They are on release version 57 now, but I need my analysis to match someone else's analysis which used release version 56.

Is there some workaround, if I can't download it anywhere? I have not been successful in finding an old mirror online yet. I have found really old versions like 45 and 8 in random places.

Thanks

[GenoMax]

Using version 57 should be fine. Refseq aims to provide stable sequence records. If you are not able to find something in release 57 then it may have been removed for a reason. You can find the records of removed accession numbers here: ftp://ftp.ncbi.nlm.nih.gov/refseq/removed/

If you must have 56, archive info is available here: ftp://ftp.ncbi.nlm.nih.gov/refseq/re...talog/archive/
Check the readme: ftp://ftp.ncbi.nlm.nih.gov/refseq/re...archive/README

[shawn.mek]

No I actually need 56, to match what was done previously.

I'm not as concerned with up to date information, I am concerned with reproducing something.

I need those missing records, and I need to not have the new records in 57.

EDIT
With the catalog file you point to, I guess I could get rid of new info from 57, but I need all the gbff info for things that are now missing as well. Plus this introduces alot of work. I'm just wondering where a previous release might be found Surely the internet has it floating around somewhere, I guess not.

[GenoMax]

Have you emailed NCBI support to check?

[shawn.mek]

Yeah, haven't heard back yet (it's been awhile), but will post later if they ever tell me anything.

Any other possible workarounds you can think of?

[GenoMax]

Any other possible workarounds you can think of?

You have already alluded to what would need to be done in EDIT part of post #3. That would be a monumental undertaking at best and can't be really considered an "official" NCBI release 56 anyway.

You obviously have a specific need for release 56 but while you are waiting for a source for 56 why not try and do the analysis against 57 and see if you can reconcile your results using the NCBI release/archive notes as to what changed between 56/57. Not knowing what kind of analysis you are trying to reproduce this suggestion may or may not be scientifically valid.

[shawn.mek]

Well, I am reproducing the analysis that used 56, but then doing additional analysis using 56 to go along with the old analysis.

If I use 57 to do the additional analysis I wouldn't have anything to compare it with. I guess I could do the old analysis with 57, but I'm not privy to exactly how that analysis was done I need to match my extended analysis to their old analysis (which was done with 56). Does that make sense?

Argg. why isn't infinite disk space free, and every file persistent forever

[8colorjino]

Hey, so did you get the older version of RefSeq release?
Can I get the link of that?

[shawn.mek]

Actually, yes they did get the older version for me and made it available temporarily. You can email them, and eventually, they will probably be able to put it somewhere for you to download. They were quite helpful.