Anybody know this company? They claim 100% accuracy but I doubt that.

[Tony0902]

I found this on rockethub. How is it possible?

http://www.rockethub.com/projects/25042-world-s-lowest-cost-genetic-sequencing

Does anyone know them? It seems like it can't be for real and they don't really explain what they are doing.

They claim 100% accuracy but how is that possible? I emailed them and they told me this:

"We have a master-mix we created which captures all the exomic and some promoter regions for hundreds of known cancer genes. The difference between our solution and the two others out there is that ours captures every base pair in the regions we target. So far we haven't missed any region of known genes we have targeted, but that doesn't mean that we won't. Worst case is that we have a little over 200 genes completed so far. Even that cancer test should be very beneficial to most people since we focused initially on the most common cancers. Plus we can sequence 20X more patients per run on Proton right now (compared to the Ampliseq Hotspot) since we do not fragment and we have figured out how to make each number of amplicons produced equivalent to every other amplicon produced in our solution, thereby reducing re-assembly and errors. And the entire reaction takes place in only 2 tubes."

I didn't understand what they were taking about at the end there but I just want to know if anyone else has heard of them or used their test. And if so, how did it work? I am always skeptical of any crowdsourcing project. I asked them to confirm the genes they already have but haven't heard back from them yet.

They also told me they would send us enough of their master-mix if we wanted to run a sample for free but i am reluctant because I don't want to waste a run on something that no one else has ever run. They claim that it replaces the Ampliseq cancer hotspot panel. Except they have complete genes instead of fragments. But they claim that they can run 20 patients on one run on the Proton instead of 1. I did the math on that. Their coverage has to be low. Or my arithmetic needs practice

OK, if anyone knows anything about this company or has run their kit please let me know. I am too skeptical right now to waste a run but would consider allocating maybe 10% of my run to compare it if they really do send me a kit for free

Thanks!

[swbarnes2]

I think those guys started another thread here.

Their refusal to use proper capitalization in their postings was not an encouraging sign.

[Bukowski]

I think those guys started another thread here.

Their refusal to use proper capitalization in their postings was not an encouraging sign.

You mean this thread:

http://seqanswers.com/forums/showthread.php?t=30343

I thought their defence of their amateurish posturing on Rockethub was brilliant.

Be prepared for a lot more of this style of promotion in the life sciences for people who want to make a fast buck.

Hint:

If you've solved one of these problems:

http://www.lifetechnologies.com/us/en/home/new-ideas/grand-challenges.html

claim your cash, don't start a Kickstarter campaign.

[math_guy]

I found this on rockethub. How is it possible?

http://www.rockethub.com/projects/25042-world-s-lowest-cost-genetic-sequencing

Does anyone know them? It seems like it can't be for real and they don't really explain what they are doing.

They claim 100% accuracy but how is that possible? I emailed them and they told me this:

"We have a master-mix we created which captures all the exomic and some promoter regions for hundreds of known cancer genes. The difference between our solution and the two others out there is that ours captures every base pair in the regions we target. So far we haven't missed any region of known genes we have targeted, but that doesn't mean that we won't. Worst case is that we have a little over 200 genes completed so far. Even that cancer test should be very beneficial to most people since we focused initially on the most common cancers. Plus we can sequence 20X more patients per run on Proton right now (compared to the Ampliseq Hotspot) since we do not fragment and we have figured out how to make each number of amplicons produced equivalent to every other amplicon produced in our solution, thereby reducing re-assembly and errors. And the entire reaction takes place in only 2 tubes."

I didn't understand what they were taking about at the end there but I just want to know if anyone else has heard of them or used their test. And if so, how did it work? I am always skeptical of any crowdsourcing project. I asked them to confirm the genes they already have but haven't heard back from them yet.

They also told me they would send us enough of their master-mix if we wanted to run a sample for free but i am reluctant because I don't want to waste a run on something that no one else has ever run. They claim that it replaces the Ampliseq cancer hotspot panel. Except they have complete genes instead of fragments. But they claim that they can run 20 patients on one run on the Proton instead of 1. I did the math on that. Their coverage has to be low. Or my arithmetic needs practice

OK, if anyone knows anything about this company or has run their kit please let me know. I am too skeptical right now to waste a run but would consider allocating maybe 10% of my run to compare it if they really do send me a kit for free

Thanks!

hi. we have actually sent our master mix out to several labs since we posted on rockethub. but only one group has actually run it yet (to my knowledge at least).

did anyone in our group talk to you about our bar-coding? that is the only way i can imagine that you would be able to use 10% of the run (unless you're not using any of the same region that we are going after). but it shouldn't be a problem.

also, we are for real. you are actually the second person that asked about us in seqanswers so i guess i am happy about that. at least some people checked us out.

other than a few labs that we have been working with and a couple of cancer studies that we are involved in we have not had very much usage of our solution yet. so i cannot tell you if it will work as we think it will. i can tell you that we have compared our data to over a hundred known cancer exomes from specific studies/patients and we haven't made a mistake yet. but we have only finished just over 200 genes so far. our goal is over 500 genes plus a bunch of regions that are related. so there is definitely a possibility that we cannot get all the genes.

i am also skeptical of crowdsourcing. i have given money to some things and never received anything. or the thing i got was not even close to what they said it would be. we don't plan to do that. even if we cannot get past these 200 genes we feel that the test is more than good enough as it is. so for most people it would still be worth it (in my opinion). and we are already at the 200 gene mark.

regarding the "arithmetic". you are actually right about the coverage. but that is how we solved this. our coverage is the same for all amplicons. the biggest difference between amplicons is about 10%. so while most solutions need hundreds of related (sonicated) fragments, we only need 40 (and we don't sonicate). and really we only need 25 or 30. plus the way we have hacked the protocol falls in line with proton's pipeline. we have run a few hundred patient samples so far and haven't missed anything yet. of course these are sample which have already been run on a hiseq at around 1000x.

let us know how it works for you. and if you have any questions.

[math_guy]

I think those guys started another thread here.

Their refusal to use proper capitalization in their postings was not an encouraging sign.

You are not the only person who complains to me about that. Sorry for anyone who gets annoyed at people who write only with lower-case. I know it is annoying. But i am so used to it now in on-line areas like forums. I feel that I write so much everyday and much of it is quick texts and emails that I am just used to it now.

Anyhow, I won't promise to stop doing that but I will try to be better

[math_guy]

You mean this thread:

http://seqanswers.com/forums/showthread.php?t=30343

I thought their defence of their amateurish posturing on Rockethub was brilliant.

Be prepared for a lot more of this style of promotion in the life sciences for people who want to make a fast buck.

Hint:

If you've solved one of these problems:

http://www.lifetechnologies.com/us/en/home/new-ideas/grand-challenges.html

claim your cash, don't start a Kickstarter campaign.

We actually did solve it. But we are not interested in a million bucks. It cost us more than that to figure it out and test it (if you count all the stuff we borrowed and all the supercomputer time we stole).

We are interested in seeing if we can hack the Proton and Miseq (Miseq is next for us) to make it give us better data. We are not interested in solving their problems for them so that they can charge us even more for data for which we are already over-paying. Oh...and we get a million dollars divided by almost 10 people...who worked almost 4 years on it. Blah.

For us it is simple: make DNA sequencing better; make it seriously accurate; and make it cheap. If we can do that then we can finally start to take the advances in cancer research out of the multi-billion dollar pharmaceutical companies and place them in smaller labs. If we could sequence 100 patients in a clinical study instead of 12 (because that was our budget and sequencing at 1000X is expensive, believe me) then small labs like our lab will be able to make actual advances instead of getting a hint and begging for more money so we can follow up on the "hint" we got from our 12-patient study.

Lastly, we never did this for money. The funny thing is this: we realized about a year ago that we were not the first group to have solved this problem (by problem, i mean seriously increasing accuracy in one run while decreasing the number of reads). We worked backwards on the pace of current cancer research as it relates to DNA sequencing and we counted the number of sequencing machines (new ones...4 years old or newer). Without question, the largest pharma companies solved it like we did (or at least close to how we did it). But they are never going to make their discovery public because they are in business to profit and not give away their discoveries. So, we decided to make it publicly available, for free. That is how little we care about making $50,000 per person or whatever (after taxes and dividing amongst our whole group). And, even if our test does nothing for the pace of cancer research, at least we tried. I am ok going out like that. ....And if, for some reason, we can make our solution work on 50,000 genes instead of 200 then maybe our group could talk about making money. ....Until then we will keep our day jobs.

[austinso]

Well if you just hacked the proton (being the self-professed math geeks that you are), my guess is that you've developed a signal processing algorithm that may be better (the Illumina system presents a very different set of challenges). But then again, you say that you have a home-brew mix...

But seriously, if you have data that demonstrates this 1000-fold (99.9% to 99.9999%...Q30 to Q60...?) improvement in base-calling accuracy, I'm curious as to why you collectively did not apply for NIH funding so that it stays within the public realm...?

If you have data that shows absolute uniform coverage across a genome, and show spike in data that corroborates your claimed sensitivity, I'm sure you'll have a lot of people interested rather than skeptical.

BTW...it would help to know how you define a 99.9999% accuracy.

[math_guy]

Well if you just hacked the proton (being the self-professed math geeks that you are), my guess is that you've developed a signal processing algorithm that may be better (the Illumina system presents a very different set of challenges). But then again, you say that you have a home-brew mix...

But seriously, if you have data that demonstrates this 1000-fold (99.9% to 99.9999%...Q30 to Q60...?) improvement in base-calling accuracy, I'm curious as to why you collectively did not apply for NIH funding so that it stays within the public realm...?

If you have data that shows absolute uniform coverage across a genome, and show spike in data that corroborates your claimed sensitivity, I'm sure you'll have a lot of people interested rather than skeptical.

BTW...it would help to know how you define a 99.9999% accuracy.

Yes to the signal processing software. But it was from that difficult attempt (which turned out to be wrong for us...kinda embarrassing but whatever) we were able to learn something else. We don't need to apply to NIH. We only needed about $60,000 total in chemicals and runs (which we "borrowed" from our labs on campus). And the super computer time we already knew we could use for free. We worked on this project for almost 4 years as a group after work and on weekends while most of us were in their Phd program or post-doc. None of our labs thought it was worth working on so we did it ourselves. I suppose our labs could end up being right...

And like I wrote earlier, we are not the only group to have done this. We are pretty sure that numerous companies are doing the same thing. I just think that maybe we're the first group that thought about trying to make it available to the public. I am pretty sure that large public prizes are there because companies are nervous that one person (or in our case, basically ten people) could come up with something that could quickly hurt their high-margin business. The problem in our case is that we don't know if it really works for anything bigger than a couple hundred genes. And as soon as a sequence vendor makes any change, whatsoever, in their software or protocol or hardware (obviously that one) or pretty much anything then we will no longer work with the new stuff. So the commercial applicability isn't really worth going after either....we would spend more than we made.

Therefore (sorry for the long post) since our master mix will be gigantic (by our standards) we thought that we would see if we could make people in the real world interested in DNA sequencing as it pertains to cancer. And see if we could use our extra master mix instead of just throwing it away as soon as the protocol is changed and the old protocol or flow cell or whatever is no longer available.

We have a lot of data (too much right now actually...wow...sequencing takes a LOT of space). But we are unable to share anything because no papers have been published (or probably even really started) yet. But I can tell anyone in this forum that our solution would only be interesting to anyone here who is using the Proton now (and if it works then the Miseq) and who is looking at the same set of genes...in other words, not many people here I would imagine. And they have to really understand how to process the master mix (it is hard, even for us right now, but that might be because we are still new at this...when some other labs do it they seem to get much better results than we do...and one time we had another lab just run it instead of us and the results were nearly twice as many useable reads as we normally have). But basically, other than our master mix and how to prepare it, we don't really have much for labs right now (I mean, it will end up costing you time and resources plus real estate on your flow cell...so...I always tell this to labs...but we will give it to any lab for free).

As for the 99.999% accuracy: once we are able to run our first public run it will be very clear to anyone who can understand the data. We will probably end up doing some kind of sort on the FASTQ data so that it is easier to understand for each sample.

We are really interested to see if the general public is interested, though. For us, we feel that is the best group for what we have. If we are lucky enough to solve the 530-ish gene kit then I would say that we have something that is worthwhile for labs. But for now I would say that labs would not be interested.

[SNPsaurus]

You should be careful about asking for funding for intellectual property that belongs to the institution that employs you as post-docs and graduate students. Any work in your broad area belongs to the employer (most likely) and you would need a license from them to carry out the Kickstarter campaign. Just a word of warning! I wouldn't want to see you lose your jobs over this.

I'm still a little confused over if your method improves the capture efficiency or the base calling accuracy. There have been a spate of papers out in the past couple years showing ways to reduce sequencing error down to the 1:5000 bp to 1:100,000 bp (Safe-SEQ, duplex sequencing, overlapping reads). You are doing something different?

[lh3]

I know little about chemistry, but I think your method (increasing accuracy by decreasing throughput) sounds reasonable. I think you should approach LifeTech who are trying everything to improve accuracy. You might get funding and meanwhile benefit more who are interested in fast and accurate genetic testing where throughput is not a huge concern.

On the other hand, I have to also agree with others: before I see the publication, or the real data, or at least some plots, I would remain cautious of your saying.

[math_guy]

You should be careful about asking for funding for intellectual property that belongs to the institution that employs you as post-docs and graduate students. Any work in your broad area belongs to the employer (most likely) and you would need a license from them to carry out the Kickstarter campaign. Just a word of warning! I wouldn't want to see you lose your jobs over this.

I'm still a little confused over if your method improves the capture efficiency or the base calling accuracy. There have been a spate of papers out in the past couple years showing ways to reduce sequencing error down to the 1:5000 bp to 1:100,000 bp (Safe-SEQ, duplex sequencing, overlapping reads). You are doing something different?

In the oddest way: it would be nice if our work did belong to the university. The UC system has no money right now and kinda needs some. But we have already spoken with our labs. We have been speaking with them for a few years now and they support what they call our "side project."

Our method improves the base calling (again...our opinion). This is about half of our solution. The other half improves the capture efficiency. That is pretty much it. Our software isn't even necessary. We only use it internally. The main part we are using "publicly" uses some of these things we worked on:

1) in our opinion we super-computed the heck out of a certain type of primer set which is better than what we had been using
2) we figured out how to get those primers to actually bind when they needed to and a specific polymerase to bind and react how we need
3) we computed and determined how to get those primers to not mess around with each other
4) we figured out a way to not use too many primers (cuz they're expensive...when you need a bunch...and when you mess up you gotta throw them ALL away and start over)
5) we figured out a way to control our reaction so that we can tell it roughly how many amplicons we need for each target we are going after)
**almost every fragment which lands and is ready for sequencing is on target...at least so far

In our lab, we combine this with our hack of their software. This gives us better results. But for what we are doing on rockethub, we are not using our software because we need the reports that we make for these people to not be based on new, unconfirmed software that we created which could have errors in it. The only change we make for that report is that we delete the primers so that they are still private to us.

[math_guy]

I know little about chemistry, but I think your method (increasing accuracy by decreasing throughput) sounds reasonable. I think you should approach LifeTech who are trying everything to improve accuracy. You might get funding and meanwhile benefit more who are interested in fast and accurate genetic testing where throughput is not a huge concern.

On the other hand, I have to also agree with others: before I see the publication, or the real data, or at least some plots, I would remain cautious of your saying.

The only thing about working with a hardware vendor, in my opinion, is that they would mostly be interested in profiting off of an improvement (not that this is wrong). But I had cancer. I am not interested in helping a company use a set of simple hacks (that won't cost them much) to increase their profits. I am interested in making the test that I took 7 years ago actually work and actually not cost so much money that only certain people can get it. Also, I am not in ANY way bad-mouthing sequence vendors. I have been thinking of a better way to sequence for a long time and when I look at these machines I am always impressed. And I am thankful that companies spend billions and billions just to design them. I just want the pace of research to increase and I decided that I would work on it part-time (which has become all the time lately ).

What we're trying to do is unconventional. I know that. We started as a couple of bio-informaticians, a guy who wrote encryption software, a guy who worked with photon detection devices, and a couple super-computing people. I ALWAYS tell anyone that we talk to that what we are working on seriously might not work. But we all got into it for personal reasons and we're mostly interested in making this test cheap and better for regular people while simultaneously cheap and better for cancer patients.

So if it were to actually work and we could convince other labs to help us out then we could probably use all of our computers' down time to compute pretty much any re-sequencing effort for any small lab and then provide them the tiny protocol changes and primers and polymerase to get so that they get the best results from ANY platform they have or if another lab has already run it then they could provide a master-mix. Eventually we would have enough master-mixes that we could run thousands of samples. For example, our master-mix for 200 genes is about 5,000 runs. we only need about 800. We can't even run than many but I would want to keep extra in case. Any lab that needed 40 of our 200 genes would still benefit, even though they wasted their real estate with the other 160 genes because the cost is lower. We pay so much for primers and current sequencing requires so many copies of the same base that it makes sequencing expensive. We just thought that a public project might speed things along...even if it were tiny.

You know...if it even works.

[austinso]

In the oddest way: it would be nice if our work did belong to the university. The UC system has no money right now and kinda needs some. But we have already spoken with our labs. We have been speaking with them for a few years now and they support what they call our "side project."

But I don't get that. Anything created with UC equipment or funding belongs to UC, does it not? The only financial barrier for UC then is submitting a provisional patent.

Unless of course, you are using technology that exists, that is in the public realm, and that is already locked down with IP, with only a "marginal" difference in execution (i.e. improved design through rigorous computational analysis), and I'm using the term "marginal" very loosely.

On paper, it is all doable, and kudos for getting it done if you actually have. Data that supports this improvement would go a long way in getting the funding your group is looking for. Problem is that what can be used in the lab cannot be used for clinical diagnostics.

[math_guy]

But I don't get that. Anything created with UC equipment or funding belongs to UC, does it not? The only financial barrier for UC then is submitting a provisional patent.

Unless of course, you are using technology that exists, that is in the public realm, and that is already locked down with IP, with only a "marginal" difference in execution (i.e. improved design through rigorous computational analysis), and I'm using the term "marginal" very loosely.

On paper, it is all doable, and kudos for getting it done if you actually have. Data that supports this improvement would go a long way in getting the funding your group is looking for. Problem is that what can be used in the lab cannot be used for clinical diagnostics.

We didn't do this with UC equipment (not too much at least). Or if we used the equipment we paid for it (like sequencing runs) with our own money or we borrowed a lane or some space on another lab's run to test something we were working on. Some of us are Phd's or post-docs from UC but not everyone. Most of us presented this to our labs but none of them were interested enough to fund it but they did help us over the years with lots of excellent advice and we traded several small projects with other labs for super-computing time. And most of our primers were traded for with a bigger project that we helped someone with.

If UC wants to submit a provisional patent then they should feel free. But they won't know what to request because our group currently owns this from start to finish, we're the only ones who understand it, we're the ones who built it, and we're the ones who worked on it after work and on weekends. Anyhow, my point is that they won't do that nor could they. We even had to pay for several full sequencing runs on miseq at a cost of $2500 for one 2x250 run out of our pocket. So....they don't even give us a discount if we can't find a lab to let us piggy-back on what they are doing.

Also, we are not really looking for funding. I know we are on a crowdsourcing site but technically we don't need the money. What we are interested in with crowdsourcing is trying to see if a consumer test would be something that people were interested in. The money we requested from our crowdsourcing effort would have been used to complete our kit so that it included everything we need for that test.

I already, oddly, made a pretty decent amount of money off of bitcoins (I know...laugh). Since i only spent $1,000 a few years ago on my bitcoins I decided that this was a decent way to use my profits on them. As of today they're up over 100 times what i paid for them. So my plan has been to use all but the original $1000 on this project. So far we still have enough to complete it (unless bitcoin drops a lot in value ).

Lastly, we believe that the kit we are building could work in clinical diagnostics. Our solution works with our software or without. We can use Proton's standard pipeline (after amplicons are created) so if Proton is ever accepted in clinical diagnostics (via FDA or whatever) then our solution would still be valid as long as our software is not used. Unless I misunderstood what you were writing?

[SNPsaurus]

austinso, I agree with you. Universities have a different take on the matter than an individual lab PI might. Here are some common rules when employed by a University:
-Disclose inventions. The university asks you to disclose all innovations, even if you don't think they have commercial value, so the technology transfer office can make that determination. This can protect you from ruining their chance of patenting an idea.
-Disclose outside activities. The university has rules about what you are allowed to do outside of your contract as a post-doc or graduate student. NIH has rules, either directly from funding a postdoctoral fellowship or from funding the lab that pays you, about outside activities. Often to have outside activities, you need to petition at the vice provost for academic affairs level.
-Disclose conflict of interest. By forming a company, there is the possibility that your academic research is tainted by your activities in the company. The university wants, and is mandated to do so, to manage this conflict by having you disclose to the public your conflict and how you are allowing oversight to prevent your academic endeavors from being corrupted.

These are taken very seriously by universities. Even worse, by "borrowing" supercomputer time (you mention millions of dollars of time for calculating primer design... sounds unlikely!) and using university lab equipment to support your commercial interests, you have taken taxpayer subsidized equipment for your company's gain. This could get very ugly if you aren't careful!