Forum:Startup in Genomics Research

Question

[Deepak]

Hello,

I am a full stack developer with a few years of experience and great interest in algorithmic coding. I am also highly interested in the field of bioinformatics and since I don't have any academic background in this field I have started taking up courses on coursera.org, reading basic biology, bioinfo books and ncbi genomics book. I am also evaluating and researching many bioinformatics startups already in the market, using opensource tools and libs.

But, I still don't have a very clear idea of what product I should build that can be really helpful to researchers, hospitals, etc.

Can some biostars please guide me with this...?

What algorithmic problems do you guys (bioinformaticians, researchers) encounter on daily basis?

Do you need beefy computing resources regularly for your problems?

How much do visualizations help with your problems? (a few examples would be appreciated).

Do you need some data discovery tools like advanced omics search engines to cater to your needs?

I know all of this would take time and I need to have patience, but if I can get some clear picture of what you guys would look for on a regular basis, I can get a good start with this.

PS : This whole post might sound a little weird but I am really hoping to startup my own company in this field and do some good research work along the way.

Thanks :)

Answer 1

[Istvan Albert]

You are looking at this from the wrong perspective: you have a solution looking for a problem. This is unlikely to be successful.

Biology is incredibly complicated and challenging, its applications likewise. Programming is much-much simpler and a whole lot easier (IMHO) and much better defined. It is very unlikely that out of the blue you could both identify a problem and produce a proper solution in a way that makes sense to a life scientist.

I would suggest that instead of striking out on your own go work for bioinformatics companies and startups for a few years and get a deeper understanding of what the problems are, what good solutions look like, what people need, and what they are willing to pay for.

[Deepak]

Hi Istvan,

Thanks for quick reply. You are right, I guess. But, I knew that about my approach and did not want to think that way :P because so far that I have seen that in established companies, s/w devs get to work on either legacy softwares or a very small part of their services which barely gives one a bird's eye view of the whole scenario. So can you give me one more piece of advice? whether I should go for more traditional approach and enroll in a university where I can deepen my understanding about problems in biology or should I look for a s/w dev job in a startup where I can understand problems with coworkers and work on solutions? What would be best approach?

Regards :)

PS: Sorry if this became more of transition from computer science to biology question. I can post this as new question if you'd like.

[Istvan Albert]

First we need to recognize that Ph.Ds are designed to prepare future scientists - not future company founders/programmers/managers. I would advise against pursuing a PhD if your ultimate goal is not research and science oriented.

I think the best bet in your case is to move to place that has a booming bioinformatics industry then take on existing jobs, get to know more and more people, situations, problems. Move around until you find the ideal place that suits you best. You need to get to know many people working the fields that interest you. There are always many many opportunities.

Answer 2

[GouthamAtla]

These are thoughts from my little knowledge on bioinformatics market and may or may not be useful for you.

When you say "evaluating and researching many bioinformatics startups already in the market", you could see how the company is structured in terms of leading scientific team.

The members will be either from strong academic backgrounds or who worked in companies for at least 2-3 years, understood the requirements of scientists and build solutions. And solutions does not need to be innovative from scratch, because most of the commercial software/companies use the publicly available tools or pipelines.

There are different kinds of companies

1. Which offer data analysis services + trainings
2. Companies which build the GUI: They build GUIs on existing algorithms and sells the software as product. The core strength of these programs is they are easy to use, but they have limited functionality and generally outdated. A good bioinformatician thinks "I could do everything in a single script what a commercial software does"**
3. Companies which builds cloud based platforms for large scale data analysis at low price and optimised the pipeline to use the computational resources in a clever way.
4. The new strategy is "Clinical Genomics" which offer panel based sequencing methods to identify predisposed diseases or rare diseases. They have tie ups with few hospitals and gets samples.
5. Others. I have no idea.

All the above models use the available algorithms. I have no idea if any company has developed a new algorithm for the needs of scientific community in genomics. For a company, investing time and resources for new algorithm design does not yield any profits.

Either you have to chose the existing model and go in that direction or create an entirely new model, both needs you to work in the field for few years. As Biology is extremely complex, it really really difficult to build a generalised solution for a problem. Each and every project has specific scientific goals and needs a specific approach.