email enquiry to BioinfoTools back to BioinfoTools home page back to BioinfoTools home page

What BioinfoTools offers

Learn about BioinfoTools’ consulting, training, software development and tools

Choosing a consultant

If you’re looking for a bioinformatics consultant or computational biologist, consider these suggestions

Contacting BioinfoTools & FAQ

Feel free to ask about your project (there is no charge for initial queries). The frequently asked questions (FAQ) is available here too.


Quick jump to: BioinfoTools Home page  >> Choosing a consultant (this page)
 

Choosing a consultant

 

Choosing a consultant can be a daunting task, especially if it is your first time. The notes below offer some points you may want to consider. The points offered here particularly relate to independent consultants (such as Dr. Jacobs) v. larger teams.

Contents

Bioinformatics is not easy

A temporary specialist member of your research team

Can you communicate easily?

Approach specialists in the planning stages

Be prepared to define the scope

Match skills with needs

For many projects, computational biology skills matter most

Look for quality control

Computer resources need not be on-site

Consulting fees

Data security

 

Bioinformatics is not easy

It is easy to be lulled into a false sense of confidence from successfully running a few BLAST searches or exploring a few tracks on Galaxy. Research bioinformatics is much more than using a few well-honed web-interfaces.

Don’t be fooled by the ready availability of software creating an optimism that your post-doc or student “will figure it out.” The time and infrastructural issues can be bigger than many realise. While it’s likely that many (if not most) will eventually figure it out—if they can commit the time involved—it’s often better use of time and money to use people that are already up the learning curve. Even simply bringing the data into appropriate formats can eat up a lot of time in the hands of those less familiar with data preparation.

There is, of course, specialist skills involved. Managing more than modest amounts of data requires experience in data management tools and/or scripting. In time, most biologists can take on this level of work, but is the time involved in learning the skills worthwhile for your project, or are you better to use someone who has already gone up that learning curve and is aware of the common pitfalls?

Bioinformatics analyses should be treated with as much respect as experimental protocols. Just as good experimental science involves a good understanding of the finer points of the experimental protocols, good computational biology requires a good understanding of the nature of the algorithms used and the theoretical biology they are founded upon. These are well out of the scope of most non-specialists. Working without fully understanding the algorithms would be equivalent to using protocols without understanding their nature or reading a research paper at face value with not consideration of the specifics of the methods used.

This issue of background knowledge becomes even more apparent when you consider each specialist area that bioinformatics can be applied, as each niche has it’s own domain knowledge. Like any area in biology background takes time to acquire and what at first may seem simple will prove to have surprisingly deep roots.

In addition to a computational biology background, if new code is required, you will want someone with sound coding experience. Good programming skills take time (years) to develop; developing new applications and the like is much more than a few lines of scripting.

[jump to table of contents]

A temporary specialist member of your research team.

A key reason to out-source a consultant is to obtain the services of an experienced scientist for a period without the need to hire and retain a full-time specialist within your group. With that in mind, you may want view the out-sourced computational biologist to be a temporary member of your research team.

Different bioinformatics consultancies differ in this respect. Some provide only “standard” analyses of data as it comes off, say, sequencing experiments. Those that operate this way may not be in a position to provide much close interaction - they may be mainly set up for running data pipeline runs with little time for working closely with their clients. Other consultancies are more able to provide close interaction over and above data analysis. (BioinfoTools does.) You should discuss if your work needs close interaction and if the consultancy is able to meaningfully provide this.

Close interaction is particularly important if you have biological questions beyond generation of initial sequencing (etc.) datasets or the analysis of the experimental data you are using is still evolving. Relying solely on data “standard” analyses is best when the data type is very well understood and their are well-accepted standard approaches.

Finally (even in the case of initial datasets), be aware of the limitations of services that only run “standard” analyses with no consideration of the specifics of your project. Each species, gene family and so on has it’s own quirks - you may find yourself having to re-analyse the data at a later stage to take into account a feature of your dataset not considered by the “standard” analysis.

[jump to table of contents]

Can you communicate easily?

You will struggle if you find you cannot communicate easily with the consultant. The consultant should know enough biology to comfortably communicate with you in a similar way as you might with an experimental biologist from another niche in your broad field. For this reason the consultant ought to be someone with a strong biological science side, ideally a computational biologist rather than someone whose focus is limited to purely the computational aspects.

In the case of services offering standard analyses from a data pipeline, you will want to see a sample of the output to see that they communicate the meanings of the results well (assuming your aims are not to have this data processed further for you by a specialist). Do they explain what each analysis does, the limitations of the methods used, clearly explain what each dataset is, and so on?

Start with the biology.

Discussion should start with the biological aims of the project. I am frequently approached with a well-meant idea of what bioinformatics is wanted to be done, and gently encourage the person to back up and describe their science (in some detail). Ideally the consultant should assist with the mapping of the biological question(s) into computational biology approaches suitable to address the questions at hand. (If you find the consultant reluctant to do this, you may be better looking elsewhere.)

[jump to table of contents]

Approach specialists in the planning stages.

Always approach those that would be doing the data analysis while planning the project, prior to writing the grant application or project proposal. They should be able to assist in writing the proposal, and make it a stronger proposal.

This cannot be over-emphasised. If you are to take a dataset to be analysed only to find it lacks in someway, at the very least it will cause a delay in the project while the issues are resolved. You will also lessen the chance of missing opportunities though not be aware of alternative analytical approaches that might be helpful. Furthermore, you may find the independent view on your project from a quite different angle helpful in writing the grant application or project proposal.

Another, pragmatic, reason to do this in the case of academic groups hiring using an external consultant is that management of the grant funds can be simpler if the consultant is named as a part of the team on the grant.

When large amounts of data are involved, researchers seem to readily look to external assistance, presumably frightened by the scale of the data, but the need for specialist advice is in many ways even more true for smaller projects that have specific questions their data is to address. The nature of the question is important than the amount of the data in many ways. (Well-established large-scale analyses often have relatively “standard” analyses compared to focused projects.)

[jump to table of contents]

Be prepared to define the scope

It will pay in the long run to define fairly carefully what is involved, both so that the likely cost can be estimated fairly and so that a reason estimate of the time the work involved can be made. Discussion of any legal, data protection or maintenance issues should take place at this stage, too.

Defining the full scope of a larger project can take time and catch the first-time contractor unawares, emphasising the need to cover this early in the planning stages of a project.

[jump to table of contents]

Match skills with needs

Look for a fit between the needs of your project and the computational biology skills on offer. This is easily the most important part of choosing a consultant; most of the other requirements are secondary to this. In the end of the day it is not the size of the consulting team that matters, but that they are able to match the skills needed for the job at hand.

Small teams or skilled independent consultants can provide good results and may in fact be a better option if the project wants a tightly focused approach, particularly if the most of the detailed work focuses on a specialist niche the consultants have particular skill in.

Likewise, big consulting teams don’t necessarily mean quality. Quality is mostly a result of a good understanding of the project, attention to detail and implementing quality control steps.

[jump to table of contents]

For many projects, computational biology skills matter most

For many projects, the biological results will hinge on the computational biology rather than particular IT technology used.

Those with solid experience in computational biology will have IT skills. By contrast those with IT backgrounds will, in general, have poor knowledge of biology at a research level. (This in part reflects that those with a background in mainly computing or computer science can find it hard to learn well the first-principles biology, chemistry and physics that underpins good computational biology. Dr. Jacobs has a joint biology and computer science background and a long association with computer technology.)

You can think of bioinformatics consulting, broadly speaking, as having four categories - computational biologists, statisticians, computer scientists (for algorithm development), IT computing specialists (for infrastructure). (Other specialists such as biophysics falling within computational biology for this purpose.) It is relatively rare to find people with deep backgrounds across the more than one of the different broad aspects of bioinformatics.

Most consultancies resolve this by hiring a mix of staff that cobbled together nominally cover the ground, facilitated by a senior consultant acting as a manager. These teams are really only as good as how well they are melded together as a team by whoever leads them. Look carefully at who is in control of the project and what time and ability they bring to the project. Even quite large groups can bring substandard efforts to the table.

By contrast, BioinfoTools offers the service of senior computational biologist directly to your project rather than junior staff via a managed approach.

Within computational biology there are many niche areas within the field as a whole. Talk to the consultant to establish what their strengths are and if these present opportunities for your project.

[jump to table of contents]

Look for quality control

Ask what testing is done, both of the software used (or developed) and of the overall methods applied to the data. Quality control is vital, particularly for processing large datasets where the outcome is uncertain. While quality control efforts are better known in sequencing projects, they applies widely through bioinformatics (although are too often neglected or done poorly).

[jump to table of contents]

Computer resources need not be on-site

Small teams don’t mean an inability to handle large datasets and a lack of large-scale computing resources on-site does not mean that you should look elsewhere. There are number of ways to access very large off-site computing resources that enable essentially any project to obtain the computing resources it needs. In addition, if the projects are academic research projects access to local computing infrastructure is possible.

[jump to table of contents]

Consulting fees

Consultants’ fees typically include all expenses. While businesses will be familiar with this, academic researchers may find it hard to compare these rates to their salaries.

To get an approximate idea, take your annual salary, add all expenses: equipment, consumables, conferences, journal subscriptions, laboratory space rental and so on. Work out what fraction of your time you are delivering output (i.e. are not at meetings, reading general background, preparing grant applications, and so on). Divide your annual salary + costs by the time spent delivering output, convert to an hourly rate and you have an equivalent to a consulting fee.

Consultants can be cost-effective once it is understood that consultants maintaining their expertise and overheads, and the typically high overheads of universities are taken into account.

[jump to table of contents]

Data security

Those with commercial projects, or academics projects using medical data or the like that have security or legal issues, should discuss these in the initial stages of enquiring about a project. Likewise software ownership or licensing fees, should there be any, should be checked. (In many cases open-access software will meet the needs of the project, in which case software rights will not be a concern.)

[jump to table of contents]