Timeline for Sandbox for Proposed Challenges
Current License: CC BY-SA 3.0
27 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 17, 2020 at 9:03 | history | edited | CommunityBot |
Commonmark migration
|
|
| Jun 9, 2017 at 16:08 | comment | added | user58826 | Hello! This looks like a good but abandoned meta post, would you be willing to offer it for adoption? (If you want to, you can still post to main.) | |
| May 6, 2015 at 21:15 | history | wiki removed | Martin EnderMod | ||
| May 3, 2015 at 15:37 | history | made wiki | Post Made Community Wiki by Martin EnderMod | ||
| Sep 4, 2014 at 21:28 | history | post merged (destination) | |||
| Aug 27, 2014 at 1:47 | history | post merged (destination) | |||
| Mar 8, 2014 at 7:04 | comment | added | Jonathan Van Matre | I'm reworking it! Have patience. I just figured I'd put up the DIM idea as well since it was less complicated. | |
| Mar 8, 2014 at 4:25 | comment | added | user10766 | Are you going to rework this or just let it die? | |
| Mar 3, 2014 at 15:41 | comment | added | Jonathan Van Matre | That is an excellent suggestion. I will be retooling this proposal toward that end. | |
| Mar 2, 2014 at 5:52 | comment | added | user10766 | Perhaps a code-challenge where people earn x points for each complexity implemented? | |
| Mar 1, 2014 at 14:33 | comment | added | Jonathan Van Matre | I am starting to think that all of these complexities should be included (this proposal stems from me noticing that most of the extant random DNA generators are pretty weak) and this should just be a popularity contest instead of a golf. Link a couple of good articles on the structure of the genetic code and let people add as many features as they wish. Making it a golf seems to be a catch-22 between too many compromises or a too-impenetrable wall of rules and conditions that will dissuade participation. | |
| Mar 1, 2014 at 14:29 | comment | added | Jonathan Van Matre | Yeah, the 20bp starting point is a bad idea. The problem with start codons is that I considered introducing the idea of promoters and decided that would make the whole thing too complex. So in the absence of promoters there has to be some way to determine which Met is the start codon vs an amino acid and the easiest simplification is to have no Mets in the gene. Likewise, for "not triplet aligned", I'm trying to avoid having to go into explanations of frameshift mutations (even though a Frameshift% would be a cool parameter). | |
| Mar 1, 2014 at 9:45 | comment | added | John Dvorak |
In fact, the 3% tolerance for the CG content leaves no room for randomness when there are only 20 base pairs. I can shuffle the pairs and turn some A<->T or C<->G, but that's it. In fact, if the CG content is set to zero, the task is impossible: we want a gene content of 2 base pairs (which is itself impossible), but the start codon contains a G, and a single G in a 2bp sequence means a 5% CG content, 2% than is the limit. Not including a gene means that we are 7% under the gene content lower limit. Similarly, it's not possible to start or stop a gene with nothing but Cs and Gs.
|
|
| Mar 1, 2014 at 9:38 | comment | added | John Dvorak |
My true random number sequence generator was sitting there watching silently as I typed away the sequence ACACACACACACAC.... It's all okay. The TRNG was capable of producing something better - it just didn't really get to it.
|
|
| Mar 1, 2014 at 9:34 | comment | added | John Dvorak |
As for the randomness, I'm not worried about the source of randomness (whatever native library is available is assumed to be good enough) but rather how the source of randomness is used (can we just start the sequence with a start codon and insert an end codon at just the right spot if it doesn't occur naturally sooner, then fill in with more random codons while avoiding ATG subsequences? Your "sufficient randomness" places constraints on the RNG (useless) but no constraints on how it's used (or that it needs to be used at all)
|
|
| Mar 1, 2014 at 9:29 | comment | added | John Dvorak |
In nature, the first ATG encodes the start of a protein coding region and defines a reading frame (triplet boundary), the rest encode methionine and the first triplet aligned stop codon encodes the end of the protein coding region (and no amino-acid).
|
|
| Mar 1, 2014 at 9:25 | comment | added | John Dvorak |
" This is acceptable, but the "gene" length in this case is calculated from the most proximal start codon to the stop codon. " - wait, what? In nature, the first one is the start codon, and the rest encode methionine. Under your scheme, methionine (which is an essential amino-acid) would be impossible to include into proteins. Your scheme would also be much harder to splice. Also, what happens to AUG substrings that are not triplet-aligned to previous AUG substrings?
|
|
| Feb 28, 2014 at 17:25 | history | edited | Jonathan Van Matre | CC BY-SA 3.0 |
added 237 characters in body
|
| Feb 28, 2014 at 17:20 | comment | added | Jonathan Van Matre | Revised accordingly, although I remain open to suggestions on how best to frame the standards for acceptable randomness. I want something that won't be exploited by answers making no effort at randomness, but that doesn't have the pain-in-the-butt factor of generating 10mb+ of data and running a Diehard test battery. | |
| Feb 28, 2014 at 17:17 | history | edited | Jonathan Van Matre | CC BY-SA 3.0 |
added 237 characters in body
|
| Feb 28, 2014 at 14:00 | comment | added | Jonathan Van Matre | Also, I've muddied the waters with RNA encoding and DNA encoding, (U vs T), which we can chalk up to a late night. | |
| Feb 28, 2014 at 13:58 | comment | added | Jonathan Van Matre | Good points. I was hoping to avoid having too much text, but that came at the expense of less clarity than the challenge demands. Edit forthcoming. | |
| Feb 28, 2014 at 12:30 | comment | added | Peter Taylor |
@JanDvorak, (part of) the point of that example is that there are two AUG substrings.
|
|
| Feb 28, 2014 at 12:11 | comment | added | John Dvorak |
@PeterTaylor AUG starts the gene, then come the CCA, UGC, CUA and GCU triplets, none of which terminate the gene. Now if there were three C's instead of two, then UAA would be the terminating triplet and the whole sequence would form a gene. I agree the definition is imprecise, though.
|
|
| Feb 28, 2014 at 12:02 | comment | added | Peter Taylor |
The only partial output example given flagrantly violates the spec. If the GC content is 42.1%, the gene content should be 31.05%, not 22.0%. The definition of "gene" is also imprecise: in the sequence AUGCCAUGCCUAGCUAA, which is the gene?
|
|
| Feb 28, 2014 at 5:54 | comment | added | John Dvorak | "Use hardcoded fragments for anything other than the start and stop codons." - why not? Specifying criteria for what counts as enough randomness should make these useless in any case. Speaking of which, you need to specify criteria for what counts as enough randomness. | |
| Feb 28, 2014 at 4:55 | history | answered | Jonathan Van Matre | CC BY-SA 3.0 |