Hacker News Reader

Comment by dekhn | Hacker News Reader

Biology of Cancer ed 2 is from 2013 and nothing has changed tremendously to the point where the book would be out of date.

Molecular Biology of the Cell ed 6 is from 2014

Molecular Biology fo the Gene is a bit older and I would generally not recommend people read it, but the problem is (IMHO) the alternative, Genetics by Lewin, is genetics-oriented instead of molecular biology, and I find that most CS folks understand MB and find genetics confusing.

Review articles are good, but I still recommend starting with textbooks before moving on to reviews.

View on HN

> Review articles are good, but I still recommend starting with textbooks before moving on to reviews.

We are agreed on that. I shouldn't have said "out of date", which suggests what is in those books is invalid. What I should have said is "they don't really cover topics that are new or controversial and mostly only cover fundamentals".

In reality as you know each new edition of those books is 95% or more identical to the previous edition.

The only downside to this approach is that if people are interested in a particular topic like cell senescence, it's a bit harsh to say "go read these 5 textbooks that may have 2 pages on the subject before you even start looking into the topic you're interested in". And the Biology of Cancer? I'm published on several cell senescence papers and don't know jack about cancer. I own the book and skimmed it though. I think Hallmarks of Cancer review is a much better introduction than this tome. The other two are good fundamentals books though. For aging I recommend Handbook of the Biology of Aging.

FWIW, I do not know a single PhD student that read those books cover-to-cover. In most of my grad school classes you weren't even required to buy the textbook and they were barely used. Actually in grad school I never bought a single textbook for classes, although I did get some for personal use.

> and mostly only cover fundamentals

I consider that a feature when entering a new field. Once I have the fundamentals, it's much easier to read a paper and have at least a rough idea if it's a crackpot theory or the real deal. Without the fundamentals? No clue.

The easy way around this is to find out what journal the article was published in and look at its impact factor. If the impact factor is below 2, ignore it. If it is between 3 and 10, it is probably legit although it may not be the majority viewpoint in the field. If it is above 10, then either the article is written by a very important person in the field, or it is considered to be potentially high-impact but may be wrong. Alternatively very massive consortium experiments are published in high-impact journals as a matter of course.

It is best to stick to the 3 to 10 range though until you know the ins and outs. Lots of Nature papers look exciting and turn out to be wrong. This is a rough guideline that varies by field.

PLoS One is a journal that publishes anything but has a decent impact factor so it is an exception to this rule. There are a few others but it generally holds true.

Thanks for providing well thought-through advice like that. Following your arguments it seems pretty clear to me that you put a lot of thought in that.

If I may ask, what's your background that tells you things like "most CS folks understand MB and find genetics confusing"?

My background is dual CS and MB. In particular, I got an undergraduate degree in Molecular Biology at UCSC and did my undergraduate thesis with David Haussler (CS professor who helped "save the human genome") on probabilistic graphical models of E.Coli genes. At that time I didn't really know much CS (just hacking), so I went into grad school (BIophysics PhD) but focused on computational chemistry, so got lots of fortran distributed computing programming experience. Based on that I was able to do a couple postdocs in bioinformatics, increasing my understanding of protein function and evolution, and then became a full-time scientist, however after several years of not being able to find funding, went to work at Google for a decade where I taught myself all the CS I didn't previously know and applied it to scientific computing. Now I work for a company that does machine learning for drug discovery.

What I've learned over the years is that genetics is a field where people who excel in abstract thinking about blobby, messy, wet objects succeed and understand the paradigms. While CS people find molecular biology, with its concerete focus on molecular entities (agents) interacting is very easy to understand. After many years of trying to understand genetics I finally realized that many of the paradigms in genetics are just useful but wrong models that don't correspond to reality.

(Not OP) Woah, thanks for the detailed answer and your perspective on this thread.

I'm very guilty of thinking "don't we have enough approximations to brute force some solutions to problems in some parts of biology yet?" and your comment have helped provide some context that these are _hard_ problems and we are still gathering information to form better models in a lot of cases.

If you dont mind me asking, where so you work? (Happy to take it to a private channel as well if you prefer )

What do you think of ancient attempts at medical models involving bio-electric energy regulating bodily functions (e.g. chi/prana)? Do you think we have a long ways to go in reconciling modern science to ancient observations?

It's considered possible to stop aging with a strict yogic lifestyle. Requires a sincere study to verify and learn from such ancient systems.

Reply on news.ycombinator.com