The TB Challenge: How should we select A, B and C segments from the larger possibilities?

General Science Discussion
71

@wuami: I’ve realized that I’m not clear on the implications of targeting the reverse complement sequences instead of the sequences themselves. If we did, would the eventual diagnostic still be submitting RNA inputs to our RNA design(s)?  Or would it be submitting cDNA as switch inputs to our RNA switches? If the latter, is there experience to tell us how much different these DNA/RNA bindings are compared to the RNA/RNA bindings we’re developing with?  If the former, what would the process be for generating the input RNA?

Perhaps my more fundamental question is “What kind of knowledge do we, as Eterna players, have that might contribute to making this decision?”

73

@omei: We’re in the process of working with the MIT Little Devices lab to figure out the exact setup for the final diagnostic device.  This will help us figure out whether we should be targeting the probe sequence or the RC.  If it’s RCs, it will be cDNAs reverse transcribed from the RNA from blood.  There’s plenty of literature on RNA:DNA hybrids out there, e.g. http://pubs.acs.org/doi/abs/10.1021/bi00035a029.

75

@wuami: if I understand the paper clearly (thanks for the pointer btw), RNA/DNA hybrids have distinct thermodynamic properties compared to homogenous RNA/RNA and DNA/DNA duplexes. And as far as I know, Nupack doesn’t support hybrids, while the Vienna team has an alpha package (2.1.6h) that hasn’t made it yet into the master branch (which has now advanced to 2.2.4).

If the choice ends up being cDNA, maybe we should go for full DNA in the Flash simulations (should be doable), using Nupack.

This said, RNA would evidently be more practical. Well, specially for me :smiley:
Joke aside, it seems to me that reverse transcription should be seen as costly, slow and/or complicated, specially for a diagnostic trying to be as cheap and as fast as possible… @wuami: please, when you find a minute to explain, I’m curious as to why this is actually being considered.

1 Like
77

@nando: cDNA is indeed much trickier than RNA.  The rationale for using cDNA would be that that is what they were actually detecting in the experiments used by Tim et al. to find the expression signature.  I should say I’m not yet convinced that that is the better option.

78

I’m still trying to wrap my head around the problem.

Nando mentioned that the 50-nt probe is just “a marker for the presence of a transcript of the human GBP5 gene. The actual mRNA (final, after transcription and splicing) is about 4000 nts long, not just a 50-nt oligo, and that’s what the diagnostic device will be dealing with.”

What I think I hear you say, is that it is not just the 3 different 50 base sequence we have to catch somehow with our switch. But those 50 base sequences each are placed inside a huge ball of messenger RNA.

Or put in other words, I have a real hard time imagining our tiny switch star fighter in between three giant death stars - and much less holding them.

Did I understand this correct, and if so, aren’t there any tiny microRNA’s we can catch instead?

80

Information gleaned in yesterday’s meeting: a reverse-transcription step would eventually allow for a selective amplification. In other words, the cDNA products would only be the signals associated with the 3 genes of interest. This would remove the worries about possible interferences from other genes.

I gotta admit: if it’s not otherwise too costly or too complicated, that’s a nice feature.

82

A couple days ago, I said:

For one, I strongly doubt that MS2 will be the signal used in the final diagnostic.

Well, as a matter of fact, we’re very seriously considering immediately using an alternative in our experiments, because MS2 “kills” the microarray chips very rapidly (and these are a bit costly…). Reporter oligos, as used in the first round of RNA-in RNA-out, look like very good candidates.

84

I think you have the right picture in your mind, yes.

The one fact we know for certain, is that a 50-nts perfect match is enough for a proper binding with one of these mRNAs. But just like you, I can’t help but having some worries about the possible steric clashes when we’ll try to have 3-4 such bindings in a pretty restricted volume. That’s a… ehm… challenge! :smiley:

1 Like
86

We are currently discussing with collaborators in the OpenTB project whether we will directly detect the RNA or instead detect small DNA or RNA segments that are output from PCR reactions that can amplify RNA from blood.

The latter amplification will make detection easier (higher concentraion of RNA to detect) and will help get around your problem of having to find the segment within a large mRNA. Its the same sort of principle that allows forensic DNA testing  or modern RNA-seq to work.

But the amplification is an extra step – our collaborators working on paper-based diagnostics are piloting how to do it cheaply and quickly.

In the meanwhile, Eterna can get us ready by solving test puzzles that use sub-sequences of the A, B, and C genes . We hope to confirm that we understand principles to rapidly get working ‘RNA calculators’ for A*B/C^2 with test A, B, and C RNAs (about 20 nt in size in Nando’s upcoming puzzle).

The resulting Eterna designs might perfectly match what we use in the OpenTB device if they choose to amplify segments of mRNA from blood into 20-nt RNA. But if they choose differently,  we might have one more puzzle using the actual DNA or RNA sequences that would be probed into the OpenTB device. I’m optimistic about success even in the first rounds of such Eterna experiments because of the player’s success on *the first round* of the A/B puzzles! 

1 Like
88

@Rhiju, thx!

I also recall reading in one of the TB papers that Wuami linked, that there were certain small molecules (Neither RNA and DNA) that were also upregulated in TB patients. I don’t recall which of the papers and what molecules. But RNA’s are great catching stuff with aptamers. Just a thought.

Wishing you and our collaborators luck with figuring what route to take.

And since you mention puzzles, I just want to highlight that Nando has put up a new  puzzle to help us prepare for the coming TB labs.

http://nando.eternadev.org/web/puzzle/3398914/

Everyone who loves switches, please give it a stab.

And there are more TB related puzzles. Those puzzles carrying an A, B and or C in their title.

http://nando.eternadev.org/web/playerpuzzles/?search=nando&switch=checked&sort=date

89

Blunt end inputs behaves different to full length biomarkers

We have one further thing to take into account, when it comes to getting a good PCR product as input for our TB concentration measuring devices. If we take the full biomarker length, we won’t get the same results as for the shorter biomarker cutouts that we use for lab. Here is why.

By doing a biomarker cutout, we can get it to make a direct coaxial stacking between reporter and input. Input and reporter like to sit right next to each other, for an ON state. This can’t be done, when there is extra bases at the end of the input that are not used. The exact same goes for designs where the full length of the reporter is not in use. The input bases and the reporter bases can’t get as close.

This is already a problem for some of our lab puzzles that makes use of an input and/or reporter, but does not use the full length of one or both ends. Such designs can never reach a KDON as low as if the input had been cut to fit perfectly for only the bases used. Or summed up:

  • No blunt end cut input, no true coaxial stacking
90

Hi Eli!  Can you elaborate, or perhaps give an example?  I don’t think I understand.

92

Sure, Omei!

I don’t think RNA like binding up, with rather sharp bends. I think that makes the input and reporter bind less sure. So if there are no excess unused bases on either input or reporter for making the bind happen, I think it they will bind better. Although a little excess bases may be a help for a faster kickoff.

Effective binding of inputs and reporter

Image https//lh5googleusercontentcom/jCw5mLNyfTv9ydOTZBZraoE6nfbaNzeHxGHP2bdFMkMI_efty3qq8Gn4UCl_CNb7LSgT33I9X_6VCMX85v_S9hTnUofo2GJCSg7YD985XiPa-zlI-DO1XvdI4ffvK8HfVPzsDBg1600×1125 414 KB

94

I was thinking today about our Tuberculosis labs and that it would be much easier if the biomarkers we used were microRNA, instead of mRNA’s.  

As mentioned earlier, I have been worried about our tiny switch star fighter in between four giant death stars size mRNA’s. 

It would be much simpler than dealing with 20 bases from a 60 biomarker from a thousand bases long mRNA, we could instead catch a small 19-24 basepair miRNA sequence - and times 4. I was thinking that this would make things easier working in blood.  

I did a search for microRNA’s use as biomarkers and they are started to getting used as such. Here is a paper: 

Extracellular microRNA: a new source of biomarkers

Here are a few excerpts: 

Based on computational prediction, it has been estimated that more than 60% of mammalian mRNAs are targeted by at least one miRNA 

This makes me wonder if there are microRNA’s for the 3 biomarkers we have been using for the TB labs?

A (GBP5)
B (DUSP3) 
C (KLF2)

While the majority of miRNAs are found intracellularly, a significant number of miRNAs have been observed outside of cells, including various body fluids [20-24]. These miRNAs are stable and show distinct expression profiles among different fluid types. Given the instability of most RNA molecules in the extracellular environment, the presence and apparent stability of miRNAs here is surprising. Serum and other body fluids are known to contain ribonucleases [25], which suggests that secreted miRNAs are likely packaged in some manner to protect them against RNase digestion. miRNAs could be shielded from degradation by packaging in lipid vesicles, in complexes with RNA-binding proteins, or both [26,27]. 

The ideal biomarker should fit a number of criteria depending on how the biomarker is to be used (Table 1). It should be accessible through non-invasive methods, specific to the disease or pathology of interest, a reliable indication of disease before clinical symptoms appear (early detection), sensitive to changes in the pathology (disease progression or therapeutic response), and easily translatable from model systems to humans. 

… secreted miRNAs have many requisite features of good biomarkers. miRNAs are stable in various bodily fluids, the sequences of most miRNAs are conserved among different species, the expression of some miRNAs is specific to tissues or biological stages, and the level of miRNAs can be easily assessed by various methods, including methods such as polymerase chain reaction (PCR), which allows for signal amplification. The changes of several miRNA levels in plasma, serum, urine, and saliva have already been associated with different diseases [39-59] (Table 2).

At last I will just mention that at least some microRNA’s have been identified as biomarkers for childhood tuberculosis: 

Circulating microRNAs as biomarkers for the early diagnosis of childhood tuberculosis infection

96

As long as the mRNA sequences that bind to the substrate are available for binding, it shouldn’t be too much of an issue. What I would be more worried about is the possibility for off-target binding between the functional RNA sensor and the mRNA. That may not actually be a bad analysis to make (similar to scoring how well a guide RNA binds to the DNA substrate for CRISPR activity in a full genome) based on the sequence space for the three gene transcripts.

98

Also something that I hadn’t thought about is the possibility for alternative splicing of the three genes. It makes me wonder which transcript variants the Khatri lab were detecting in their assay, or if it’s a mixture of all transcript splicing variants.

100

I partially redact the first statement. It’s possible (and I vaguely recall this from a seminar I attended) that the overall negative charge of the mRNA, which is quite high, may lead to a form of aggregate separation when combining these larger molecules with a bunch of smaller particles. It’s partially based on the concentration of the components, so yeah, the larger molecules could definitely crowd the smaller functional RNA.

With that said, the method of mRNA amplification is probably using two primers to amplify a specific region of the gene. Granted, that’s more a question for the Das lab, and I’m sure these are concepts they’re taking into consideration at this time.