Why do SHAPE data look different for the same design, when viewed in the lab list versus as design?

Hyphema found a real oddity in the lab results of one design. A barcode that didn’t fold as expected though it really ought to have folded. Nando pointed out that the data looked quite different in the lab list instead. There it looked fine.

Images of the exact same design, shown in lab list and as design with barcode area marked in both:



Nando mentioned different threshold as a bid on what could be wrong and I am thinking something with legacy scaling showing up again in the design interface, where it ought to be absolute scaling.

So what is wrong?

We need the data to be identical in both interfaces.

Background chat:
Hyphema: eli, i had something i wanted to show you from a lab but i cant seem to find it… [10:37 PM]
Hyphema: http://eterna.cmu.edu/sites/default/f… [10:47 PM]
Eli Fisker: Ok, tell me about it, Hyph. I see one real odd thing [10:48 PM]
Eli Fisker: But you first [10:48 PM]
Hyphema: heres a pic without the shape scores [10:48 PM]
Hyphema: http://eterna.cmu.edu/sites/default/f… [10:48 PM]
Hyphema: i dont get why with an overall good structure here that the barcode shape essentially sucks [10:49 PM]
Eli Fisker: Good point [10:49 PM]
Hyphema: should have been a 100 [10:49 PM]
Eli Fisker: The two images look the same [10:49 PM]
Eli Fisker: I don’t get why the barcode suck [10:50 PM]
Eli Fisker: because it looks like one that would normally do fine [10:50 PM]
Eli Fisker: 4 GC pairs that are nicely distributed [10:50 PM]
Eli Fisker: It really should have hold better [10:50 PM]
Hyphema: http://eterna.cmu.edu/sites/default/f… [10:52 PM]
Hyphema: this is jandersonlee’s mod of a vinnie design scored 98 [10:53 PM]
Hyphema: looks great. barcode although different than mine but is simmilar [10:53 PM]
Eli Fisker: http://prntscr.com/52foqu [10:54 PM]
Eli Fisker: Same barcode, doing well in a different design [10:54 PM]
Eli Fisker: Though a few of this barcodes identical siblings did less well: http://prntscr.com/52fp4r [10:55 PM]
Eli Fisker: Ok, so your and JL’s design are identical except barcode? [10:56 PM]
Hyphema: no, just making a point that in the same round a winner had a good barcode shape score. therefore the round was not necessarily frought with errors [10:57 PM]
Hyphema: i just dont have an explanation with my barcode [10:57 PM]
Eli Fisker: Ah [10:58 PM]
Hyphema: seems to me nando’s auto barcode script based on a dotplot (i think) would have chosen my barcode as acceptable. however it didnt turn out that way [11:00 PM]
Eli Fisker: Your lab design is level 2 or one in the bulge cross lab? [11:00 PM]
Hyphema: http://prntscr.com/52frx5 heres my dot plot [11:00 PM]
Eli Fisker: Found it [11:01 PM]
Eli Fisker: And yes, that looks pristine [11:01 PM]
Eli Fisker: I wonder if we can find similar barcodes in the cross lab [11:02 PM]
Eli Fisker: Because this one has better error rates than the bulge cross, though this is in the better end compared with the other history tour labs [11:03 PM]
Eli Fisker: And I understand why you wonder [11:03 PM]
Hyphema: the bulged cross lab modified gap 2 [11:03 PM]
Eli Fisker: Yep [11:03 PM]
Nando: @hyph: have you checked how that barcode looks like in the web data browser? http://eternagame.org/web/browse/4835… [11:03 PM]
Hyphema: what do you mean. but i guess not [11:04 PM]
Nando: check it out [11:05 PM]
Eli Fisker: Now I really wonder [11:06 PM]
Eli Fisker: Why are they so different? [11:06 PM]
Nando: I would guess that it has to do with thresholds… [11:06 PM]
Eli Fisker: Are design showing legacy scaling and lab list showing absolute scaling? [11:07 PM]
Eli Fisker: I mean lab list and design should be identical [11:07 PM]
Eli Fisker: Else it makes no sense [11:07 PM]
Hyphema: this stinks. but i have got to go. please continue to chat and i will hpefully read your thoughts on chat later. afk [11:08 PM]
Nando: I haven’t followed the story about legacy/absolute, so I couldn’t tell [11:08 PM]
Nando: bye Hyph [11:09 PM]
Eli Fisker: Hyph, thx for bringing this up. I think I will throw it up in the forum [11:09 PM]
Eli Fisker: It do stink [11:09 PM]
Eli Fisker: http://prntscr.com/52fv5b [11:09 PM]
Eli Fisker: http://prntscr.com/52fvr3 [11:09 PM]
Eli Fisker: Identical area marked but very different [11:09 PM]
Eli Fisker: Bye Hyph [11:10 PM]


The SHAPE threshold (that splits the blue/yellow color scale) is set to the “legacy” scaling as it has been identified previously.


It is pretty easy to view the reactivity measurements in the Eterna interface, through the advanced option of raw data. The data is identical to what is uploaded into the server and displayed in the web browser.

Most likely, since the threshold is computed with every sequence, the code does not have an override for the threshold values being uploaded at the same time.