The Meenta.io Blog

Cloud Sequencing Platform

[meenta-io]

Three Mistakes In Searching for NGS Service Providers

Here are our favorites. If you think they have merit then read further;

1. Using search criteria that is too narrow by defaulting to local resources.

2. Using search criteria that is too broad by using Google. 

3. Type II errors.

Search Criteria Too Narrow

We know, you love your local core (or it’s a love hate relationship?) or were told by your institution that you HAD to love them. In either case, no post-doc, grad student or PI has time to evaluate or validate another NGS service provider, let alone a full comparison. The technology is changing so fast we would have to revalidate them every 6-12 months… out of what research budget and whose time? Fact is most scientists (~60%) do use their local core sequencing facilities out of sheer convenience. However, most users are unsatisfied with their experience mostly due to the lack of timeliness of getting on the right instrument coupled withe the uncertainty of when data would be returned. Ask yourself, how many service providers or core labs did you vet before sending them your samples? If you said more than 3 you can go have a beer and kick back.  Otherwise keep reading.  Why isn’t there a way to compare multiple offerings and determine quality without having to meet with the facility staff, send multiple e-mails and the dreaded quote process for pricing?

The time sink associated with vetting a new next-generation sequencing facility forces scientists to default to a “favorite” NGS facility. We find this odd as most bench scientists know about the issues with false negatives, yet ignore it when it comes to selecting the services they use. Stated in another way, just because there is no evidence of something better, is there none?

The Problem With Search Engines:

We all know it! We have all become too complacent and almost “co-dependent” on relying on search engines or social media or Alexa as the extent of the universe!  We turn to all these tools for information and sometimes validation.  While clearly not comprehensive this reliance has infected science as well.  It’s not clear to me how it has happened that well-trained scientists trust a search tool that was never designed to find scientific related things! I am talking about Google.  I am sure you have done it and typed; “NGS service providers (your city)” and received a result such as this; http://www.ngsmassachusetts.com or http://seqanswers.com/forums/showthread.php?t=948. How disappointing! Now what? You have to click on EVERY link, search every offering on THAT site, ask for a quote and then compare them manually (off-line), make a decision and then send the sample?  We would never buy a pair of shoes this way and would actually dismiss sites that made us shop in this way as low quality or as poor customer service. Yet as you may have read in my prior blog (“Three reasons not to buy another sequencer.”) we have come to accept this pain in our scientific lives.  These search engines only provide lists of things, but nothing truly actionable as scientists would expect. THIS is the very reason companies like Airbnb and Amazon exist…they are easy to access, fully transparent, AND you can act on them immediately, no waiting.  Stop using search engines and use a dedicated tool!

Type II Errors

Alright, let’s confess we don’t remember what a type II error is (https://en.wikipedia.org/wiki/Statistical_hypothesis_testing)!  Our normal null hypothesis with using cores or service providers is that if a colleague made the recommendation that it must be better than others you have thought of.  This is absolutely not the case since chances are that your colleague has just offered a qualitative judgment with no data or validation to support their reference. Not because they don’t care, but because it’s been a long time since they did or they simply don’t want to spend the money on it.  The only way to avoid this error is to run well characterized control samples at the facility in question, which may be unrealistic for many reasons. 

Take away: Look for services that allows you to compare facilities without the need for quotes, e-mails and guarantees data turn around time. Checkout the NGS map at map.meent.io

[meenta-io]

Your Next-Gen Sequencing Map

Part One of a Three Part Series.

Think of this map as a platform for collaborations. De facto it also happens to be your spring board to finding Illumina NGS instrumentation. Please bookmark this page as the map will evolve daily!

Three key features to keep in mind;

Find instruments through out the US. 

Get alerts for instruments as they become available.

Make recommendations for new instruments or locations.

Add your instruments.

Last, a word of recognition to others who have worked on this topic in the past such as  James Hadfield (Cancer Research, Cambridge, UK) and Nick Loman (University of Birmingham, UK). 

We aim to extend on their efforts and humbly thank them for their contributions to date to the NGS community!

[meenta-io]

The Empty Lane Problem

by Gabor Bethlendy

The ‘Empty Lane Problem’ is an understated issue in the Next-Generation Sequencing (NGS) space with a big cost/sample impact. It turns out that NGS economics is very sensitive to capacity utilization.  The same principle that has rapidly reduced the cost of sequencing – parallel processing of samples – works against you if you don’t fully load a run.

At Meenta, we think of lanes on sequencers like seats on an airplane.

Financial sensitivity to capacity utilization is found in almost every industry. In the airline industry, for example, with its high fixed costs, airlines want planes to operate with as many passengers at as high a price as possible.  But there is a trade off between demand and price (ignoring a few other considerations such as schedule).  Fly a route too often and the airline can’t profitably fill enough seats.

Airlines manage this problem in different ways depending on the options available to passengers.  On popular routes served by multiple airlines passengers can shop for prices, departure times and seating options of their choice.  On these routes airlines deploy logisticians with fancy computer algorithms to optimize flight frequency vs. filled seats at various prices to make up their schedules. The optimum mix is probably some complex version of a Marginal Revenue = Marginal Cost graph from Economics 101 with some factor figured in for profit tolerance. On captive routes serviced by only one or few airlines passengers are at the mercy of the airline schedule with minimal options for time or price. On these routes airlines make sure they have enough passengers at a desired price point by creating an inventory of passengers to fill seats.  They make you wait.  The trick on these routes for airlines is not to create so much excess demand as to entice competition.  

Services like Priceline and Orbitz have built their business models around these problems: discounting pricing to fill unsold seats with algorithms to adjust pricing in real time.

Sequencing labs operate like a captive airline route: they wait to run a sequencer until they have samples for all the lanes.  Every core lab is different, but wait times can vary from weeks to months.  Price/cost economics is more important that the researchers/clinicians inconvenience or time.  The obvious difference between airplanes and sequencers is that airlines can move their equipment (airplanes) to meet demand where sequencers are generally stuck in a location - bit paradoxical that the gazillion pound machine moves and the almost portable machine doesn’t...

The solution is to provide a live marketplace for core labs to fill unused lanes with samples.  

Both users and sequencing facilities benefit. With live, lane-level access to sequencers, Researchers and funding providers can now factor the individual price of lane into their grant and experiment budgets. For researchers, the experiment design and budgeting can focus on short term changes in their experiments, and a per lane price. Researchers can break experiments down, and process samples across multiple trusted core labs.

For Core Labs and large research institutions, the process of filling lanes to maximize instrument utilization becomes much easier, decreasing queues and wait times. Cores can factor higher utilization numbers into their pricing.  

Everyone benefits.

This is the goal: change the underlying assumptions and economics of equipment ownership. The result is faster and cheaper science.

The Complexity Problem: Matching Samples To Sequencing Lanes

Running an Illumina HiSeq with 2 of its 8 lanes empty is more impractical for a core lab than it is for an airline company. For science and research this problem is described as the ‘Empty Lane Problem’, where the passengers are the samples or lanes on a sequencer. Unlike the travel industry, which can afford to build in some unsold seats on a flight, equipment owners are much more restricted in their options.

Equipment owners have few techniques for adjusting to unsold or missing lanes. In practice core labs solve this issue by building up a de-facto sample queue. They assume their instrument will have downtime because it can not be run half full, so factor the cost of downtime in their per sample price. And when demand is too variable, the core lab simply do not buy more equipment.

While pricing and equipment utilization is important in the grand scheme, the impact of missing lanes impacts researchers and their project on a daily basis. When a researcher comes to a lab with samples they need to run on a given instrument, the core lab manager needs to fit their request into the core labs queue of samples, both on a first come/first serve basis, but also on a equipment and configuration basis. The question is often, do I have any samples in the queue that can be combined with what this researcher needs. If the answer is yes, then core lab tells both the waiting sample owner and the new sample owner when the sample will run. If the answer is no, then the core lab puts the sample in its queue, waits for a match, and tells both researchers to wait.

The more variations of lanes and configurations in the queue, the harder it is to fill lanes and schedule the sample to run. The impact of this issue is that samples wait in equipment queues, researchers wait for data and equipment stands unused. Waiting drives up costs and slows down research. While research equipment was never purchased or priced with any assumption of 100% utilization, the impact of missing lanes impacts every lab and research facility.

Here is a Tweet-able summary… “How The Empty Lane Problem Increases The Cost Of Next-generation Sequencing.”

[meenta-io]

THE GENOMICS BREAD LINE

Author: Gabor Bethlendy

I grew up in communist Hungary and moved to the United States with my family at the age of eight. How we got here is an interesting story better left for another article. The point is that most kids growing up in the US had chores like taking out garbage or mowing lawns. However, in Hungary my chore was to ride my bike to the only store in town and stand in the bread line, sometimes for hours. My reward for this was often a Coke I bought at the bar next door. The year was 1975 and for my brother and me  standing in a bread line was a normal part of life with hours spent trying to game the system to get the best bread. 

The Bread Problem

Hungary had lots of bakers, many really good. But the bread had to navigate a difficult path from the baker to my family’s tummies: Few trucks, poor storage, and too few stores to distribute.  There was too much bread for some areas and too little for others; many loafs grew stale or moldy or went wasted, uneaten. We never knew what we would find.  But often when I arrived to the front of the bread line I was amazed to find that the shelves were full!  How could this be when there seemed to be plenty of workers handing out bread? 

The Sequencing Problem

When we visit academic or commercial sequencing facilities we see the same issues. Plenty of lab technicians, plenty of instruments, unused machine time and plenty of empty, unused lanes on sequencing runs.  Yet in many labs, freezers are full of partially filled boxes of various types of prepped libraries…all waiting to get on one or another instrument at some yet to be determined date. Why is this?

This is genomics’ equivalent of the breadline, folks with advanced degrees waiting for data - science’s equivalent of bread - with little control over access or timing. What a waste…

It is somewhat ironic that after 40 years in the US I find myself starting a company focused on solving a distribution and access problem, not for bread but for genomics. Making bread and sequencing samples are actually very similar businesses and as such manifest very similar problems. Both are abundantly available, yet both resist the notion they are “commodities.” Both require a machine - an oven or a sequencer - a few simple ingredients and well-trained operators who all claim they are the best at sour dough or Chip Seq. However, being skilled in production doesn’t translate to matching market needs.  Nobody gains if the bread goes bad or the lane goes unused no matter how delicious the loaf or good the prepped library is.  

Meenta Solves Genomics’ Breadline Problem!

Cloud based solutions can put an end to Genomics Breadlines. No, we don’t mean a Google search that results in a list of service providers! Those lists are not actionable and don’t speed up or decrease the cost of science. We propose an app that will tell us the availability and status of every single scientific instrument (@MeentaNGS) in the world.  We mean shared access with the ability to book time on them! We mean optimizing machine utilization by matching samples to available machine time throughout the world. Until then we will have to know by tribal knowledge which shops have good bread and which good library preps. With our app one can identify the good bread without the lines!

 Here is a Tweet-able summary… “Don’t Get Caught in the Genomics Bread Line!”

[meenta-io]

Three reasons NOT to buy another sequencer

Author: Gabor Bethlendy

Below are the top reasons. If you think they have merit then read further.

1. Reagent “discount” economics don’t add up.

2. There are already too many next generation sequencers (NGSs) in the world already, and buying another under-utilized sequencer is a business model that favors the NGS manufacturers, not the facilities who use them.

3. More instruments does not equate with more access.

Let’s start with #1: Reagent “discount” economics don’t add up

First, the high price of, and poor access to, NGSs combine to drive up their service fees.  Second, the full price of the instruments is being amortized across the life of the instrument (typically three years)—when only 48% of its capacity is being used during this period (48% is the median utilization rate)! These amortization costs are also wrapped into genomic service fees.  

So what do the clever sequencing vendors do?  They offer you a second or third sequencer for a large reagent discount arguing this will bring down your per sample cost and hence the cost of your services when in reality all of those costs have already been captured in their calculation of the instrument price itself!

This is a false economy, or put more simply, there’s no such thing as a free lunch…

Which leads to my next point…

#2  There is already too much sequencing capacity, and more sequencers merely serves the NGS vendor’s interests

According to the latest figures, Illumina accounted for nearly 80% of the market share of next generation sequencers and about 4,800 instruments globally. By my calculations, this is >1 peta base-pairs of global capacity (to keep the argument simple, we’ll ignore the other NGS providers like Life Tech, Pac Bio and Oxford).

This is enough to sequence about 1 million genomes at 30x coverage on one of the Hi Seq instruments.   At the moment there are nowhere near 1 million genomes being sequenced each year and genome sequencing only represents a small (~10%) fraction of the NGS use case.  So why are institutions buying more of them?

One reason may be the above-mentioned 48% median instrument utilization rate. So while theoretical capacity may be high, demand out strips capacity due to poor instrument utilization.

Figure: Worldwide sequencing overcapacity is reaching ridiculous levels

We are often lead to believe that core facilities and sequencers are “busy” because we hear (or experience first hand) how long it takes to get on an instrument and get data back.  The erroneous conclusion we reach is to buy more sequencers to get better access, faster access or better data quality.

This is not the full picture. Yes, we often hear core facility managers argue that this is because NGS is a “very complex” workflow or that it is “not a commodity” and therefore difficult to scale up.  And while I agree in part, the fact is that academic and even many commercial labs are very bad at making full use of the sequencers they already have.  Either they don’t have enough business or more likely that their mission is to make discoveries not to make things more efficient. 

However, these are all truisms that distract us from the huge problem of poor access to expensive capital equipment in all of Life Sciences! The problems are efficiency and access.  And, the fact that these concepts aren’t as a sexy as that paper in Nature or Science is used to full advantage by the sequencing vendors; they are quite happy with the status quo as it keeps us naive scientists writing grants with taxpayer dollars merely to improve NGS vendor profits.  

Which leads me to the last main reason not to buy another sequencer…

#3: It’s all about sequencing access!

By having us buy more instruments for a larger reagent discount the vendor has done a Jedi mind trick on an entire industry all to give us “better access” and “some savings.” And as we discussed above, this is a false economy.

So, is the main problem the cost of sequencing or better access?  

I know many of you would say; “but I have a core or I send my samples to a collaborator or I search Google” and it’s easy for me to get on a sequencer.  And I think this is what most people said before Uber and ZipCar…that it is easy to hail a taxi as they are all over the city.  Yet, today Boston alone has about 40,000 Ubers (not to mention Lyft) circling the city on a daily basis!  How is this possible when there are enough cars and public transportation available already?  

Uber and ZipCar have lead consumers to an important understanding…for most, transportation is a service, and like many services these days, it is often less expensive and more convenient to rent it when we need it.

I am an American, and having my own car was really tied into my notion of American freedom and independence. But there is one thing I am even more attached to—my money.  That and anything that can lead to less headaches or more time in my life.  The convenience of ordering (and not owning) transportation has certainly helped in this regard.

Today, in the Life Sciences we are just like many car owners before Uber and ZipCar…we do not yet appreciate how difficult it is to get access to sequencers and other instrumentation…nor do envision how convenient it can be!

We have accepted the pain of how it is today.  

We have accepted that we send it to the local core or our favorite service provider.

So here’s my question: Assuming quality were the same, would you care which sequencer access you use? One selected through an Airbnb-like application or would you rather wait for your “favorite” one like you do today?

So here are your choices; buy more capacity (NGSs)?  send out? or rent Sequencing Access-as-a Service (SAaaS) in the cloud?

[meenta-io]

We are in Beta+

Author: Stephan Smith

Actually we are in a bit more then Beta. Since our platform is live! But beta gives us the cover to collect more user feedback and roll out feature changes.

Remember, Google kept gmail in beta for years, so maybe we will too!

Meenta is an instrument sharing service for science, initially focused on Next Generation Sequencing (NGS). Every year researchers and research organizations invest in expensive equipment from vendors like Illumina, ThermoFish Scientific and Pacific Biosciences. These instrument are complex to house, service and operate.

More importantly they often sit unused, something as much as 60% of the time. Our mission is to unlock these investments and give labs a way to speed up their sample queues, re-coop their investments.