Hacker News new | past | comments | ask | show | jobs | submit login

I have spent 11 years in molecular biology research.

I must warn you about a survivorship bias. All the exciting news you are hearing are the tip of a giant iceberg that rests on the mounts of a most dull and repetitive labor.

Here are some realities that nobody tells you:

1. There are only 3 operations done in experimental biology: liquid pipetting, opening/closing tubes and moving tubes between machines. 90% of protocols may be reduced to those operations.

2. Just to reproduce an already published work, you need 6 month of 10 hour/day work, performing those 3 operations from point 1. It took 5 years to reproduce just 18 molecular biology papers [https://www.the-scientist.com/news-opinion/effort-to-reprodu...], while 1 out of 4 computer science papers can be reproduced in less than 30 minutes [http://reproducibility.cs.arizona.edu/].

3. It takes days, if not weeks, to perform an experiment and make an observation. That is, if you manage to make a mistake in your experiment at day 2, you will only learn that something went wrong at day 6. Often, it is even impossible to tell what went wrong.

3a. Imagine, that after days of coding, you hit "compile", and the compiler would find a typo you made somewhere in your code, and instead of pointing at the error, it would simply delete all your code, and never tell you what was the reason, so you have to start over writing the code from scratch. This is how day-to-day biology work looks like.

4. CRISPR technology is one of the most heavy on 3 operations from point 1.

5. If you ask biologists about how do they feel about doing experiments, 97 out of 100 will say they hate it the most and curse the day they decided to do biology.

Regarding starting a company, I would first try to talk to as many as possible industry people, trying to understand what they do day to day and what they find most problematic, difficult or annoying to do, and see what you need to do to solve their problem.

In fact, I so much hate the fact, that I spent 15 years studying, then 5 years doing PhD, just so I can spent the rest of my days pipetting liquid. I decided to try to automate away all manual work by creating a universal robotic framework. If you are interested, please check https://cartesianrobotics.xyz/




I was coming into this thread to recommend looking for entry level part time mcb/bio lab work precisely because of point #1.

It's a wax on, wax off thing. The repetition gives you a chance to reflect on the rationale for your technique and bigger picture of what you're doing, in addition to getting hands-on experience in the field.


I think a part-time job would be an overkill

I have trained many students to do biology, and I can teach anybody, without any education, to perform molecular biology experiments in 2 weeks.

I 2 months, they are usually able to generate scientifically valid ideas for research (although, usually not the best ideas).

Biology is extremely simple as a field, and a computer scientist will have no problem getting into the field (while opposite is usually hard)


I'd say there's no subtitute for hands on experience. In the same way people on here recommend writing your own project to learn a language over simply reading about it, actually doing lab work gives you an understanding that no biology textbook can provide.

I vehemently disagree with the claim that biology is a simple field, and I'm especially surprised to hear this from someone who did work in this field for 11 years. Anyone who has taken an intro biochem class would back me up on this: life is complicated!


I completely agree about hands-on experience: yes, it is extremely useful to practically do a biology engineering project. What I claim, is that about 2 weeks to 2 months of that is enough.

Regarding your second paragraph: I was fortunate to work in 3 fields in my scientific career: organic chemistry, materials science/organic semiconductors and molecular (synthetic) biology. This gave me ability to compare. What I found out, is that biology has the most interesting problems nowadays, and biology day-to-day work is extremely, outrageously, ridiculously stupid. I can not even start describing how amazingly debilitating molecular biology methods are. Right now, in the middle of the night I am stuck in the lab doing a day-and-half long protocol, all consisting on steps like "put liquid in", "pipette liquid out", "incubate for 5 minutes" and "shake for half hour".

Now here I am, after 15 years of school, 5 years of PhD and 11 years after that, doing work which is less intellectual than flipping burgers in McDonalds.

Fortunately, the fact that biology is so simple, gives me hope that it can be automated fairly easily, which I am planning to concentrate on for the next 5 years.

Please don't mix the plain complexity of life, which appears like "too many things to memorize", and complexity of mathematics and computer science, which is "understanding deep layers of abstractions". While I haven't taken intro bio classes, I read books and tons of papers. They are just plainly simple, you don't need brain to read them, you only need to memorize.


Genetics and evolutionary ecology falls into the category of "layers of abstraction." Mechanisms for up and down-regulation, self-catalysis, epigenetic.

You're convoluting the simplicity of each step of an experiment with the level of complexity of the field itself. Memorization is a crucial component, but if you believe that's all there then you are severely underestimating the study of life. It would be like claiming that since CS can be boiled down to 0's and 1's that it is therefore a simple field of study.


Good luck with the robot. I wonder, if the individual steps are really that simple, why hasn´t it already be automated already?

And if you don´t mind me asking, what´s the best way to start a little home experiment to see if one may actually be interested in biotech?


Thanks! The reason it was not automated yet, is that non-biologists (including CS engineers) don't know that individual steps are that simple; while biologists do not have experience in thinking in abstractions, and so they simply not seeing that they are doing effectively same operations, just with different reagents.

In general, biologists are very busy pipetting faster than their peers, so they can secure scarce professorship positions. They simply have no time taking step back and looking at more general picture.

People who have skills necessary to build a robot and understand biology, have opportunity to work for companies that pay salaries 5 times higher than academia pays, so they don't bother.

I would not recommend doing any home experiments, especially without professional laboratory experience. Most of the experiments contain dangerous, poisonous or potentially explosive chemicals, and as for a newcomer, the risk is too high.

Instead, I would recommend finding a local biotech interest group, join them and go from there. Alternatively, you can find a local biotech lab and volunteer to help a grad student or postdoc with their project. This is more common than anyone may think. This way, there are no obligations, and you can leave if you don't like it; at the same time, you will get exposure to the state of the art biology.

Again, don't be afraid that you will not be able to bring anything good; biology as a science is ridiculously simple, and if you already have CS or electrical engineering background, you will have no trouble catching up within a month or two (that is how long it takes to our grad students who come from non-biology background to start bringing in new valid scientific ideas).


There are no wet lab communities out here (Switzerland), at least I couldn't find any. But asking a postdoc to help out for free sounds promising.

Btw, I work 50 hours a week plus travel, I don't think it's going to be that fast to get up to speed.

Cheers


Interesting. It seems automation would greatly speed up this field.

I wonder if progress is being made on this front.


You are absolutely right.

After those 11 years in molecular biology, I am absolutely convinced that the bottleneck of the field is not some fancy method, but a total lack of automation.

Modern "CRISPR" cool methods are done with essentially the same tools as the research from 50 years ago.

I would claim, that computer scientists are lucky ones: they have skills to create their own tools. To make a better, more automated programming language, you would need same skills as to use that language.

Biologists on the other hands are different. The skills to do biology are totally different from skills needed to automate biology. And so the gap appears: biologists don't even think of automating, and mechanical/electrical/computer engineers are not even aware of the problem




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: