saurabh20n

Ah, sorry. Happy to explain any pieces in more detail. I hope the core is clear: If you think of Dapps as servers, then it synthesizes its clients (specifically ones that cause assertion failures).

For bugs: The demo videos on the FAQ page show the kinds of bugs it can find, e.g., the one that crashed the DAO, and ones that simpler tools will be unable to find. Of course you are right, the holy grail is finding bugs that escape both human audits and other tools. Haven't run it on very large codebases yet. Working on it!

10-50 lines of code. You have to pick the right domain. Sumit is doing a lot of work [1] on integrating these ideas into Microsoft's products, e.g., Excel's flash fill. Ras [2] is synthesizing biology models and architecture design. Sanjit, Armando, Emina, Rishabh, are all doing excellent work getting them to scale.

[1] https://www.microsoft.com/en-us/research/video/flash-fill-fl...

[2] https://www.frontiersin.org/articles/10.3389/fbioe.2014.0007...

Actual work towards this: http://sparkz.org/ai/program-synthesis/2017/10/11/neural-pro...

There are few comments here: https://www.reddit.com/r/ethereum/comments/2l0ht4/alternativ..., but seems unlikely that was the entirety of the discussion.

Agree. Deep learning does not bring us closer to AGI. It might get us closer to other proxies of "mechanical intelligence" that will be very productive.

I now believe we are 3 years from building an AI that writes Python well enough to build itself, based on some experiments I did recently: http://sparkz.org/ai/program-synthesis/2017/10/12/self-hosti...

Most technical people will understand the difference between programming and AGI. The general public might not.

The useful thing out of AGI discussions, is that they engage the general public.

Very interesting. I work in AI. Curious about trends in your industry. Please ping me at the email address in my profile.

#22 - "system should write code that sorts a list, rather than just being able to sort lists."

Using symbolic techniques, synthesizing algorithms is tractable. See Sec5.3 on pg11- of http://saurabh-srivastava.com/pubs/popl10-synthesis.pdf - Figure 2(b) is the synthesized selection sort algorithm, and Table 3 has QuickSort synthesis time at 160seconds. Very old work (2010) and used templates to guide synthesis. SAT solvers have improved, and much more CPU cycles are available, so in 2016, exhaustively exploring the template space will be trivial.

I.e., rather than reconstructing code from concrete program execution traces, giving the synthesizer (AI?) the ability to manipulate symbolic spaces leads to a tractable search space. Finding specification-matching programs within this symbolic space is doable.

One possibility is as an extension of the untargeted analysis: run the analysis over different kinds of samples. The output for each sample is the list of major peaks (and intensities). Use this as the "image" to train a (shallow) network.

You might even get away without specifying pre-identified chemicals. Adding that list would only help.

Agree that this is not written with the rigor of a journal paper. Our intention was to communicate the simple wins we've had employing deep learning in comparison to the tools of the prior generation. XCMS is the most widely used library: http://www.bioconductor.org/packages/release/bioc/html/xcms..... It requires very painful parameter tuning. Internally, we had also built our own custom targeted analysis. In the targeted pipeline, we had to pre-specify "acetaminophen, shikimate, chorismate...". After building this deep learning workflow, we have exclusively switched over to it: no chemicals pre-specified, no parameter tuning. With about 185 engineered yeast that need analysis, with replicates, feeding conditions, and controls, these simplifications have been helpful.

We are getting easy wins over microbial data. Human data is noisier and we're testing over that now. More later.

If you have microbial data, or have used XCMS in the past and would like to compare, happy to chat. email me at saurabh@20n.

20n | San Francisco | Full Time, ONSITE | http://20n.com | Genetic engineering, Machine learning, Data Mining, and Bioinformatics

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce materials, pharmaceuticals, and other products. As our first demo, we created an organism that eats sugar and produces Paracetamol -- the active ingredient of Tylenol. This is the first biological/sustainable/eco-friendly means of making it.

20n has existing Fortune 500 customers with great incoming revenue. We are also backed by Khosla Ventures, YCombinator, and DARPA. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes at lab-in-the-clouds. We are looking for experts in genetic engineering, machine learning, and algorithms, who want to engage in new ways to manipulate and use biology!

Full job descriptions:

* Scientist, Microbial (strain) engineering: http://20n.com/jobs.html#team:make

* Machine learning: http://20n.com/jobs.html#team:predict

* Data mining: http://20n.com/jobs.html#team:input

* Bioinformatics: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, Data Mining, Bioinformatics, Microbial engineering

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce pharmaceuticals, materials and other products. We created an organism that eats sugar and produces Paracetamol (acetaminophen) -- the active ingredient of Tylenol. This is the first biological means of making paracetamol.

20n has existing Fortune 500 customers with great incoming revenue. We are also backed by Khosla Ventures, YCombinator, and DARPA. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes at lab-in-the-clouds. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Scientist, Microbial (strain) engineering: http://20n.com/jobs.html#team:make

* Machine learning: http://20n.com/jobs.html#team:predict

* Data mining: http://20n.com/jobs.html#team:input

* Bioinformatics: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, Data Mining, Bioinformatics, Microbial engineering

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce pharmaceuticals, materials and other products. We created organisms that eat sugar and produce acetaminophen -- the active ingredient of Tylenol. This is the first biological means of making acetaminophen.

20n has existing Fortune 500 customers with great incoming revenue. We are also backed by Khosla Ventures, YCombinator, and DARPA. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Microbial engineering scientist: http://20n.com/jobs.html#team:make

* Machine learning: http://20n.com/jobs.html#team:predict

* Data mining: http://20n.com/jobs.html#team:input

* Bioinformatics: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, Data Mining and Bioinformatics

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce pharmaceuticals, materials and other products. We created organisms that eat sugar and produce acetaminophen -- the active ingredient of Tylenol. This is the first biological means of making acetaminophen.

20n has existing Fortune 500 customers with great incoming revenue. We are also backed by Khosla Ventures, YCombinator, and DARPA. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Bioinformatics: http://20n.com/jobs.html#team:sequence

* Machine learning: http://20n.com/jobs.html#team:predict

* Data mining: http://20n.com/jobs.html#team:input

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, data mining, NLP, and computational biology

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce valuable chemicals. We created a yeast that eats sugar and produces acetaminophen -- the active ingredient of Tylenol. ps: Our cells are the only living cells known that produce acetaminophen.

20n is a YCombinator, DARPA, and Khosla Ventures backed, with existing Fortune 500 customers and incoming revenue. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Data mining and NLP: http://20n.com/jobs.html#team:input

* Machine learning: http://20n.com/jobs.html#team:predict

* Computational Biology: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, data mining, NLP, and computational biology

Biology is severely under-utilized. 20n (YC W15) fixes that by letting software predictions direct genetic engineering. Our system modified a baker's yeast, and it now "brews" acetaminophen -- the active ingredient of Tylenol. ps: This is the only living cell on the planet that produces acetaminophen.

20n is YCombinator, DARPA, and Khosla Ventures backed, with existing Fortune 500 customers and is already profitable. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that modify the cells. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Data mining and NLP: http://20n.com/jobs.html#team:input

* Machine learning: http://20n.com/jobs.html#team:predict

* Computational Biology: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, data mining, NLP, and computational biology

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce valuable chemicals. We created a bacteria that eats sugar and produces acetaminophen -- the active ingredient of Tylenol. ps: This is the only living cell on the planet that produces acetaminophen.

20n is YCombinator, DARPA, and Khosla Ventures backed, with existing Fortune 500 customers and incoming revenue. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Data mining and NLP: http://20n.com/jobs.html#team:input

* Machine learning: http://20n.com/jobs.html#team:predict

* Computational Biology: http://20n.com/jobs.html#team:sequence

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, data mining, NLP, and computational biology

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce valuable chemicals. We created a bacteria that eats sugar and produces acetaminophen -- the active ingredient of Tylenol. ps: This is the only living cell on the planet that produces acetaminophen.

20n is a YCombinator, DARPA, and Khosla Ventures backed, with existing Fortune 500 customers and incoming revenue. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Data mining and NLP: http://20n.com/jobs.html#team:input

* Machine learning: http://20n.com/jobs.html#team:predict

* Computational Biology: http://20n.com/jobs.html#team:sequence

Indeed, YCombinator gave us the $120K.

Our total funding + revenues are much higher. You missed the part about Khosla Ventures, DARPA, and Fortune 500 contracts, and that we are already profitable :). Biology, especially when done through software, is a very profitable domain.

If the candidate is right, yes. Email us at: jobs@20n.com

20n | San Francisco | Full Time, ONSITE | http://20n.com | Machine learning, data mining, NLP, and computational biology

Biology is severely under-utilized. 20n (YC W15) fixes that using software predictions to create cells that produce valuable chemicals. We created a bacteria that eats sugar and produces acetaminophen -- the active ingredient of Tylenol. ps: This is the only living cell on the planet that produces acetaminophen.

20n is a YCombinator, DARPA, and Khosla Ventures backed, with existing Fortune 500 customers and incoming revenue. Our core technology is a data mining and machine learning platform for biological data. We routinely distribute using Spark, write approximation algorithms for NP-complete problems, and push the software predictions to robots that build the microbes. We are looking for algorithms experts who want to see their predictions create new life!

Full job descriptions:

* Data mining and NLP: http://20n.com/jobs.html#team:input

* Machine learning: http://20n.com/jobs.html#team:predict

* Computational Biology: http://20n.com/jobs.html#team:sequence

email me: info@20n.com

Indeed! We love this community [1]. We started 20n with DARPA's help, in part to gather more steam around the accuracy of software predictions tools within synthetic biology. You seem to have worked on these tools, and so must know the pain of convincing wet-lab scientists to follow up on the sw predictions.

Paracetamol, both for academics and to pharma, was a non-biosynthesizable molecule. But once we had the prediction, we were able to go to the lab to construct the microbe pretty easily.

In the end, the tool is the start of the process. We are investing significant resources in constructing the microbes. Over the coarse of the next two year, hopefully you will find some novel molecules in the ones we move to bioproduction.

[1] Have to! We are part of it. :) We know most of them, and they probably know us.

Yes. When going after non-trivial chemicals the economics work out, and is cheaper when scaled up. The process essentially repurposes beer fermentation. The usual benefits of biological production: carbon-negative, and not using any toxic catalysts or producing waste products hold as well. Consider the molecule, paracetamol/acetaminophen/Tylenol [1], that we created a microbe for: The US alone produces 35000+ tonnes of it, all currently coming from petro-routes. It would be nicer to get it the same way we get beer, if we could.

There might be additional market forces that make a microbial fermentation attractive: e.g., in the case of the anti-malarial drug artemisinin, fluctuation in supply of the plant Artemisia annua, the price of the drug varied between $120-$1200/kg, and Amyris and Sanofi moved it to yeast based production for creating a steady supply [2]. That route was critical for the supply of the drug to African countries.

For many of the requests we get, chemical synthesis is not economically feasible to take the chemical to market, and bioproduction might be the only route. The fact that we suggest routes that work well for the planet is a good side-effect. :)

[1] http://en.wikipedia.org/wiki/Paracetamol#Synthesis

[2] http://en.wikipedia.org/wiki/Artemisinin#Synthesis_in_engine...

Hi glowing plants :)

Indeed. We like our simple to manipulate (relatively speaking) microbes, but would be happy to see what avenues there are in other customer domains.

Hah! Oh PKS-es! Megasynthases (PKS, NRPS, FAS) do not currently feature very heavily in our algorithms, but their modeling is certainly within our intended algorithmic expansion plans. As for whether they are understated/overstated, my cofounder can sit you down for hours and speak about them, but you have probably moved on!

ps: Googling me will lead you to program synthesis: programs that program programs :) which is another curiosity that you may or may not want to dive into. You will find a lot more biology from my cofounder. google scholar: chris anderson synthetic biology.

Completely understand your sentiment. I have been releasing software as open source for 10+ years -- all of my 7 years of program synthesis PhD sw is available as open source.

With 20n, we had to tread more carefully. As monocasa implied, open source here means either code, or dna data. While code stays on the computing platform, dna's platform is every living entity.

Lets assume you could predict precisely what a piece of DNA will do -- we are not there yet, but will be at some time in the future. Are you advocating that all of that should be freely available? There are lots of benefits to open source and we understand a lot of it in the context of code, but IMHO, we will need to discuss more before we fully comprehend what it means to open source dna.

There are tons of opportunities, e.g., enabling bioproduction, and hopefully the beneficial nature of that (cheaper/carbon negative/environmentally-friendly) will be the leading argument in the conversation than anything else.

Tech people. Ideally, somebody with algorithms chops, but one who doesn't get scared by biology or chemistry.

Defensibility for us: microbes we build, and the platform they form. Our software-driven engineering is significantly faster than the state-of-the-art: human's looking at metabolic maps. People will start building on top of our initial microbes and improving them. These initial microbes form the "platform" from which to enable more bioproduction apps.

Many limitations! Constructing the designs (i.e., going from data->dna->cells) takes time and is expensive: we pay vendors for the construction. So we have to restrict our software to output designs that yield the most novel products, with the least change to the biology of the microbe.

Most of the time changes do nothing! But if our models are accurate enough, the engineering can be predictable. We are continuously improving our models -- from more data mining, and from feedback from experiments.

That said, what our algorithm predicts right now is still a huge space (5000-10000 products). That is the tree image in the article. For context, the chemical industry centers around 70,000 products.

Optimizing for large scale production is indeed what most of the field works on. There have been many success stories: an anti-malarial drug through yeast (Artemisinin by Amyris+Sanofi), a plastic precursor through bacteria (1,4-BDO by Genomatica+DuPont), and more.

What they are missing is the whole spectrum of what could be made biologically. We will create microbes for the most valuable chemicals and then partner with existing optimization companies that have industrial fermenters running. Think beer fermentation, just instead of the alcohol yeast, you use our yeast.