Faced with an empty, freshly painted lab room, many young scientists have a small panic attack: What do they need to properly start their lab? And how do they keep it running smoothly? To address some of the questions floating around in young investigators' minds, Genome Technology caught up with a few experts to get their take on setting up a lab. With their advice, the experience can be less stressful and more like being a kid in a candy store.
The following experts kindly took the time to offer advice for young investigators for everything from buying instruments to stocking consumables to acquiring clinical samples.
• Lynda Chin, Dana-Farber Cancer Institute
• Aedin Culhane, Dana-Farber Cancer Institute
• Christopher Cullis, Case Western Reserve University
• Bettie Graham, NHGRI
• Ivo Gut, CEA/Centre National de Génotypage
• Nate Lakey, Orion Genomics
• Gregory May, National Center for Genome Resources
• Josh Mendell, Johns Hopkins University
• Akhilesh Pandey, Johns Hopkins University
What follows are excerpts of our interviews with them.
What funding opportunities are there at NHGRI for young investigators?
Bettie Graham, NHGRI program officer: One of them is the K99/R00 which is the Pathway to Independence Award.
The K awards, which are mentored awards, give people more training and research in the interest of a particular institute. The K01 is one of the ones that we use the most as a way to recruit individuals with degrees in physics, chemistry, math, computer, and engineering sciences into genomics. That's more of a training mechanism with some degree of independence toward the end — we strongly recommend they submit an application by the time their five years are up.
One thing that [NHGRI does] do to encourage new investigators is that we try to make allowances for them since they are new to the system [and] their applications may not be so favorably reviewed. [If the application] has all the good qualities in it, we do try to fund those applications and we also try to fund them for up to five years, if that's what they are requesting. We try not to cut the years that we fund them, at least in the beginning. We also do that for people who are coming in, they may be R01 grantees in their first five years, [for their first] continuation, we try to encourage them as well. There are results to show that that's a critical point in people's careers. Many people don't make it to the second renewal, so we try to encourage that.
Are there grants specific for purchasing instruments?
BG: The National Center for Research Resources has equipment grants. In general, most research grants will allow you to purchase equipment. I think the only times there might be problems [are] where you may be purchasing a big piece of equipment that you may not be using 100 percent of the time, but other people in your division can use it. That's not the kind of thing that a regular research grant would support. That's what the -National Center for Research Resources would support.
Is there any other advice that you would offer to young investigators?
BG: The biggest piece of advice that I can give to anybody is for them to contact the program director at an institute at the concept development stage of their application.
Yesterday I had to deal with a situation where a young person submitted a small grant application to us dealing with an area — it has some genomics in it, but it was clearly cancer-related. We don't support that because our mission is developing resources, technology development. When it comes to diseases, that's not something that we support. Technically, it belonged with the cancer institute, but the cancer institute didn't support the R03. We had to end up sending that application back. In other words, this person has now wasted her time writing an application, her institution put effort in it, and then it gets here [and] the subject is not appropriate for us. The mechanism is, but not the subject. The institution that should do it, the topic is of interest, but the mechanism isn't.
I said, "Did you talk to anyone?" And she said, "No, I assumed that since I was using genomic techniques, that it would be appropriate for NHGRI." I just said, "Unfortunately, that was a bad assumption. In the future, the first thing you do once you have an idea that you want supported is to talk to somebody." Because I am sure that had she spoken to somebody, they would have counseled her differently.
The one [piece of] advice that I always give when talking to people is to contact the program director — talk to them about what you are interested in [before submitting a proposal].
How do you decide whether to use a core or service lab versus having the instrument in your own lab?
Christopher Cullis, Case Western Reserve University: [Ask yourself] what does it take to run it — and that's more specifically in terms of time. If you're a young investigator just starting out and you have an instrument that requires some maintenance or assistance in running it and you don't have a technician, then you have to do it. That is another big, big time commitment. If it's a really big-ticket item, other people are probably going to want to use it [and you're] going to have to run samples for them or going to have to do some training or have some oversight for it.
What should investigators take into consideration when deciding which gene expression analysis platform to use?
Lynda Chin, Dana-Farber Cancer Institute: The scientific question should dictate what tools to use. I encourage the new PI to take a step away from what they are familiar with technically and formulate a scientific question of interest, then design the right experiments. Based on what the experiments are and what types of results are expected, select the right tools, including gene expression analysis platform or whatever else.
Aedin Culhane, Dana-Farber Cancer Institute: Though previous studies should not restrict future work, it is sometimes useful to investigate if there are related in-house or published studies that might be useful in validation of results. If there is a related study with a large sample size, it maybe useful to consider potential cross-platform analysis issues. Other considerations are the availability of platforms in local core facilities and expertise of the bioinformatics/biostatistics colleagues that will be analyzing the data. The is especially important if considering ChIP-seq or other new technologies, as these may take longer to analyze and, thus, it may be longer before a lab gets their results.
Josh Mendell, Johns Hopkins University: For standard gene expression profiling experiments, I suggest choosing the platform based on the expertise available at the investigator's institution. It helps tremendously to have a core facility or a collaborator who can assist with experimental design and interpretation of results.
How do you choose a sequencing technology?
Gregory May, National Center for Genome Resources: We had originally looked at all three [existing technologies, from 454, Illumina, and Applied Biosystems]. We did at that time a bake-off — essentially, we gave the same nucleic acid sample to all three groups, and we compared the data we got back. And we also compared the workflow and the cost per data point.
And for us and our small research group, the Illumina platform was our best choice at the time. All these groups have continued to make improvements. After we had our first instrument, we realized quickly, 'This is really working well in our hands, let's get another one,' and the demand was there. And part of continuing with the same platform is just ease of management of not [having] two different platforms, two different supply chains, technicians trained on two different platforms and instruments.
(Interview excerpted from InSequence, March 10, 2009.)
Ivo Gut, CEA/Centre National de Génotypage: I used to put up this slide in presentations, where I showed three cars, a Formula One race car, a rally car, and a tractor. And then I would say, 'Decide which is the best car.' And then I'd show a Formula One track, the picture of a desert, and a potato field. And then I'd say 'Now think again about what you thought a moment ago,' because it completely changes. A Formula One car on a potato field is going to be good for absolutely nothing. It's really that you have to think what your problem is, and you have to choose the right tool to solve your problem.
For example, when you try to do whole-genome sequencing, the GS FLX can give you a little bit of added tweaking of your final sequence. You don't have to generate huge coverage with it, just very little. And then we think that some applications might actually do better on the SOLiD instrument than on the Illumina instrument. There is one particular problem that I'm thinking of, and that is the directionality of RNA. Applied Biosystems is telling us that they have solved the problem of providing the directionality of the RNA molecules in RNA-seq, and this is a very critical issue. Regarding error rate, in RNA-seq experiments, the error rate is not such a big player. It does play into de novo sequencing versus resequencing.
But it looks like the read length of the SOLiD is going to be shorter than what you get from the Illumina instrument, and read length is important. We are now members of the International Cancer Genome Consortium, and to us, being able to read 100 bases rather than 40 bases is critical.
(Interview excerpted from InSequence, January 27, 2009.)
Tips on buying a big-ticket instrument?
CC: Talk to the reps and find out what they're offering, and then ask them for the name of somebody who's using their instrument. They'll always give you someone who's happy with it, but even people who are happy with it can tell you, 'Yeah, I'm happpy with it, but it did take me a year to get it set up.'
How do you develop standard operating procedures for a lab?
LC: If a long-term technology platform is required, I strongly suggest recruiting dedicated scientists/technologists. Trainees (students/postdocs) are not good at this and this type of experiments is not a good training project.
CC: My initial recommendation always is make sure, whatever package you start up with, you have a full-time technician. That's especially true for faculty — they can't get in the lab [during the first year]. They all think they can, but they can't. In order to get set up and running, they've got to have someone who's there all the time. That person can help train new graduate students too. Most faculty who stutter, I think that's a big reason for it.
JM: I have found that it helps tremendously to have a responsible and motivated lab manager who can act as your "right hand" in establishing standard protocols and more generally a productive lab culture.
How should scientists plan their consumable use and budget?
Nate Lakey, CEO, Orion Genomics: There are two balancing forces. One is the cost of running out of reagent and losing time because people in your lab are idle. The term for that is called a "stock out." [The other side is] what does it cost you to buy a reagent and hold it? In this age where much of the labor costs are avoided by really fantastic kits that have done a lot of the work for us, moreso now than ever the PI may need fewer people but [the costs are more tied up in consumables].
There are a couple of simple things to do about this. We tend to overstock our fridges in laboratories. It's not necessary for a scientist to inventory every single thing they might need. The cost of overstocking is pretty high — a lot of these reagents aren't stable in the long run. Look at what the average weekly requirements are going to be for each item that you want to inventory, and also calculate what your reorder point should be — that level of inventory where when you hit that level, you reorder more of the item. Some weeks you might not use any [reagent], some weeks you might use four times the amount you normally use, [so it's important to know] your standard deviation.
If researchers do a good job with this they won't have gobs of cash tied up in inventory that's in the fridge, so if their experimental plans change they won't have all these antiquated kits and their money's still in the form of cash, not in reagents. The goal is to have as little inventory as you need without suffering substantial stock-out costs.
Another thing for someone who's starting is to build a real strong rapport with their peers and to never be afraid to walk next door and ask for some flour. Often you'll want to try something and it doesn't work — in that case you're better off owing someone a cup of flour.
Bridging the wet lab/dry lab divide
What is your lab's approach?
Akhilesh Pandey, Johns Hopkins University: Our lab's approach is that you should always do bioinformatics in conjunction with wet lab experiments. Our lab is basically a fusion of bioinformatics and wet lab in the same lab, not just under the same roof or the same building, but within the lab. We have bioinformatics people embedded in the wet lab people and also most of our bioinformatics people have some sort of a biology background. They may have a master's degree in biology or a bachelor's degree, but then they have gone on to do more programming or database work. They actually have more appreciation for what goes on while they are doing the computational aspects. I also encourage them to try their hand at proteomics experiments. We have also succeeded in having some people to do both.
About how much time does your lab spend on each aspect of your research?
AP: I think of the people in the bioinformatic institute I started in Bangalore as part of my extended lab. There are about 40 people there and I have about 11 people at Hopkins. At Hopkins I would say 30 percent is computational and in India, maybe I would say about half of it is computation and database and half of it today is experimental. They do gene expression microarrays, more higher-throughput experiments and proteomics.
It is a seven-year-old institute and the first four years, [we] only did informatics and I thought that was not a reasonable way. The only way we could sustain also the interest of people and get them to progress in their careers is by having experiments. So now in the same building, under the same roof, just like my lab except it's bigger, they do 50/50 and I am pushing very hard to basically have more and more experimental work going on there. That doesn't mean people have to leave computation. It means they can now do experiments in addition to what they are doing.
Bridging the basic/clinical divide
Do you do clinical research?
AP: One of our goals is to find biomarkers for disease and cancer, so we are working on many different areas there. Most of our work, I would say, has to revolve around human medicine. How can we make it better? I take "biomedical researcher" very seriously — that the "medical" part has to be important. I think that we have a limited time on the planet and it's my belief that many of the lessons we might learn from model organisms, although many of them are very, very popular, you might not be able to find ways to connect them easily to the human scenario. We work on human and mouse for the most part. We do other things as they relate to medicine. That means that we have to have samples of many different types that we are analyzing, from cancer to urine.
How do you obtain clinical samples for your lab?
LC: Approach that person with a question that is of scientific interest to the collaborators as well. Engage them on a scientific level. Do not treat or consider them as "sample provider" alone.
AC: Clinical samples are frequently limited. Access to fresh-frozen tissue can be difficult, so consider if the analysis can be performed using paraffin embedded tissue (e.g. DASL arrays). Consider the amount of tissue and the percentage tumor cell to stromal/non-tumor cell tolerated by your assay. If so, consider if you need to engage a pathologist to dissect the tissue. In addition to the translational aspect of the research, these are questions that are likely to be important. It is worth analyzing publicly available gene expression data to provide pilot results to support your hypothesis, if possible. Resources, such as the gene expression atlas, oncomine, or protein immuno-histochemical staining from the human protein atlas database, are all easy to query. Clinical investigators will be more interested to share valuable samples and allocate resources to your project if you can demonstrate its importance.
JM: Of course this largely depends on the culture of the institution, but I have found that clinical collaborators are very approachable, and these collaborations are highly mutually beneficial. I believe that it is important to provide a good rationale and openly share all preliminary data with a potential collaborator. Try to get them as excited about your project as you are.
AP: I am actually a trained pathologist, but because we are so involved in research and many things require focus, I pretty much do full-time research now. The way is to have collaborations and deep collaborations because I think there's still a big divide. I hear this also among similar faculty members where they actually don't like it, all this focus on translational research.
But for people who are younger and don't have any set attitudes, I think it should be a given. I see actually nothing wrong in us trying to get something back in terms of improvement in human health in the shorter term as opposed to a really long term — 20, 40, 60 years. We can try to have some tangible objective to be covered in five to 10 years.
Unfortunately, the clinicians are also very busy and they don't have the time to keep up with advances on the basic side. That means the person who is going to do all the legwork, almost all of it, will be the basic science researcher. But in a way, the problem will have to be defined in many ways by the clinical field. I think that is an issue — even when a basic science researcher picks up a problem, they may pick up a problem that maybe is not so clinically relevant. They pick a problem, but it doesn't need to be solved.
At most academic medical centers, there is more and more cross-talk, but there is still domain issues. People don't want to let go. They want to keep control of samples and things like that. But if one persists, then I don't think it's a major problem.
More specifically in the field of proteomics, we could really argue that many of the mass spectrometry groups that would call themselves in the physics, chemistry types of a few years ago, they are applying [their energy] to certain kinds of samples. Maybe they have only gotten to the point where they are looking at Drosophila cells, but you can see that application has become important — and sooner or later, they will come to analyzing things that are more important.