if this helps those of you who can do overlays well
It would be a good idea to let folks know about this option from the start. I will post in chat and Discord.
You can load a solution from 1820, but you can't use it for "load as guide".
I'll look into why "load as guide" isn't working.
Why would the development team allow players to load builds from puzzle 1820 into this puzzle? Are you looking for refinements to those prior builds, or do you really want us to start with the structure you have presented?
Wouldn't it tend to stifle the competition if I load my high scoring build from 1820?
I added this info to the puzzle description.
This starting topology is just a suggestion, as we have no idea how correct/incorrect it is.
However, we can imagine players modifying their previous predictions based on this model... or even creating hybrids between the two (which is why I want to fix the "load as guide" issue).
Sorry if this is a naive question; is it known whether the first 15 AA are a signal peptide that is cleaved off in the actual protein?
Since we do not know the structure of this protein, we can only speculate.
I did run it through a Signal Peptide Prediction Server, SignalP-5.0, and indeed that was the prediction:
The starting model provided looks just like the one PSPRED indicated, I was wondering why it looked so similar to my last attempt... http://bioinf.cs.ucl.ac.uk/psipred/&uuid=3c238016-7823-11ea-b4c4-00163e100d53
It is likely that most people used the PSIPRED predictions we posted for the last puzzle:
Since this puzzle has the same sequence as in Puzzle 1820:
should we still expect none of its 7 cysteines to form
disulfide bonds? Should we also expect all 7 of its
cysteines to be on the protein exterior?
As we don't know the structure of this protein, we really can only speculate.
I did run it through a Disulfide Bonding State Prediction Server, DISULFIND, and indeed it predicts 3 disulfide bridges (with 58.9% confidence):
I was to reiterate, however, that all these tools: DISULFIND, SignalP-5.0, and even PSIPRED, are all just statistical predictions... they are often correct, but are sometimes very wrong.
Proteins tend to follow their own rules, and they love creating exceptions (I learned this the hard way when studying knotted proteins for my PhD! :-)
In line with the 25,83 37,90 topology, some article on bioRxiv says this is gonna be an Ig fold:
No idea about 61,102 though. My hunch is that it is possible, since this pair only occurs in the SARSr-CoV insertion.
If the N-terminus is a signal peptide and if it is cleaved, wouldn't it be useful to post the puzzle at some point without the first 15aa? Even if there is no cleavage, the N-term will be in the membrane, so in Foldit any solution separating that helix from the rest of the fold will score horribly and players will be less likely to follow solutions like that.
That is a very good suggestion, Wilm.
We do have to keep in mind that we cannot be 100% sure that the N-terminus is a signal peptide, it's not like we have a cryo-EM map to work off of, we just have the sequence.
We'll definitely consider posting a trimmed version of it, if we end up running another puzzle for this protein... as there are many SARS-2-CoV protein sequences with no experimental structure available.
The Diamond Light Source in the UK has switched into emergency mode - if somebody can crystallize the ORF8 with and wit out the signal peptide and has good ties to DLS we could get fast turnover XRC