6 ms·

I dropped out of a PhD program because I was disillusioned and never found a path of value to my group's work. After reading this I really wonder if what was mi

by siavosh 15y ago

I dropped out of a PhD program because I was disillusioned and never found a path of value to my group's work. After reading this I really wonder if what was missing was context and world experience. So much of what's wrong with a lot of academia is group think and insular publish/perish mind set. What would make me want to go back at some point in my future is putting research into a larger context and grounded in my own experiences of what is truly important and valuable.

jseliger 15y ago
After reading this I really wonder if what was missing was context and world experience. "'Science is a wonderful thing if one does not have to earn one's living at it.' -- Albert Einstein" -- from Philip Greenspun's Women In Science: http://philip.greenspun.com/careers/women-in-science http://philip.greenspun.com/careers/women-in-science . I would guess the Apple VP has enough money to never have to worry about money, which probably makes the grad school experience quite different than yours.
- siavosh 15y ago
  I think this can be very true of theoretical research, but when I entered grad school I was pretty interested in the applied fields. My problem was I didn't have substantial industry experience to identify the labs that were doing real applied work compared to the publishing factories that drew tenuous claims to the real world to get grants.
  - xtacy 15y ago
    Not many graduate students have really substantial industry experience when they start. Internships help a lot, provided the goal is to learn (and not be pressurised to publish papers even there. :)) It was probably unfortunate that you didn't like your experience with your group; but not all groups are like that. It depends on your advisor.
Estragon 15y ago
I'm on the verge of dropping out of bioinformatics, too, after about a decade of postdoc appointments. The corruption of biomedical research by funding imperatives [1] is heart breaking. That said, cancer genomics could be well the field which makes bioinformatics clinically relevant. [1] http://www.labtimes.org/labtimes/issues/lt2011/lt02/lt_2011_02_24_31.pdf http://www.labtimes.org/labtimes/issues/lt2011/lt02/lt_2011_...
- dirtyaura 15y ago
  Do you think that bioinformatics is over-hyped as a field? What are the biggest issues preventing it to become relevant?
  - Estragon 15y ago
    I should have been more specific. I was mostly thinking of sequence analysis like the PhD candidate in the OP wants to do. Bioinformatics in general covers a lot of ground which is directly clinical in nature, like image analysis. Sequence analysis has had an important role to play in genome annotation and comparison, but the recent (extremely expensive) attempts to apply it to the genetics of human disease have been quite disappointing so far. The biggest technical issues are firstly that our tools for measuring genotypes are too imprecise and too expensive to support experiments which will nail down exactly which variants are contributing to a given disease and secondly that almost all related studies of human disease so far are epidemiological in nature, because you can't do experiments on actual humans (and I'm not arguing that you should be able to. :-) This means it is very difficult to establish a variant as causative, and hard to design reliably repeatable experiments. The political issues arise out of health research funding being fundamentally motivated by hope and fear, leading to massive commitments to research strategies which are wishfully ignorant of the above technical issues. This is followed by a lot of wishful ignorance of the strategies' failures by the groups who implemented those commitments. Usually this ignorance comes from moving the goal posts post hoc. The current sunset of GWAS is a good example. (See Peter Visscher's article in the January issue of AJHG for example.) The technical issues are being worked out. Better genotyping will be developed, and at least some disease biology can be modeled in animals or cultured tissue (cancer might be good for this.) But those issues have been ignored for too long, at tremendous cost, because the serious funding rewards such ignorance. For instance, essentially the same strategy as GWAS is now being pursued on a huge scale, just using a different genotyping technology. No one has run an initial small-scale experiment validating this new approach, it's just based on the hope that the extra information provided by the new technology will provide the ingredient that the studies to date have been missing.

by siavosh 15y ago

jseliger 15y ago

After reading this I really wonder if what was missing was context and world experience. "'Science is a wonderful thing if one does not have to earn one's living at it.' -- Albert Einstein" -- from Philip Greenspun's Women In Science: http://philip.greenspun.com/careers/women-in-science http://philip.greenspun.com/careers/women-in-science . I would guess the Apple VP has enough money to never have to worry about money, which probably makes the grad school experience quite different than yours.

siavosh 15y ago
I think this can be very true of theoretical research, but when I entered grad school I was pretty interested in the applied fields. My problem was I didn't have substantial industry experience to identify the labs that were doing real applied work compared to the publishing factories that drew tenuous claims to the real world to get grants.
- xtacy 15y ago
  Not many graduate students have really substantial industry experience when they start. Internships help a lot, provided the goal is to learn (and not be pressurised to publish papers even there. :)) It was probably unfortunate that you didn't like your experience with your group; but not all groups are like that. It depends on your advisor.

Estragon 15y ago

I'm on the verge of dropping out of bioinformatics, too, after about a decade of postdoc appointments. The corruption of biomedical research by funding imperatives [1] is heart breaking. That said, cancer genomics could be well the field which makes bioinformatics clinically relevant. [1] http://www.labtimes.org/labtimes/issues/lt2011/lt02/lt_2011_02_24_31.pdf http://www.labtimes.org/labtimes/issues/lt2011/lt02/lt_2011_...

dirtyaura 15y ago
Do you think that bioinformatics is over-hyped as a field? What are the biggest issues preventing it to become relevant?
- Estragon 15y ago
  I should have been more specific. I was mostly thinking of sequence analysis like the PhD candidate in the OP wants to do. Bioinformatics in general covers a lot of ground which is directly clinical in nature, like image analysis. Sequence analysis has had an important role to play in genome annotation and comparison, but the recent (extremely expensive) attempts to apply it to the genetics of human disease have been quite disappointing so far. The biggest technical issues are firstly that our tools for measuring genotypes are too imprecise and too expensive to support experiments which will nail down exactly which variants are contributing to a given disease and secondly that almost all related studies of human disease so far are epidemiological in nature, because you can't do experiments on actual humans (and I'm not arguing that you should be able to. :-) This means it is very difficult to establish a variant as causative, and hard to design reliably repeatable experiments. The political issues arise out of health research funding being fundamentally motivated by hope and fear, leading to massive commitments to research strategies which are wishfully ignorant of the above technical issues. This is followed by a lot of wishful ignorance of the strategies' failures by the groups who implemented those commitments. Usually this ignorance comes from moving the goal posts post hoc. The current sunset of GWAS is a good example. (See Peter Visscher's article in the January issue of AJHG for example.) The technical issues are being worked out. Better genotyping will be developed, and at least some disease biology can be modeled in animals or cultured tissue (cancer might be good for this.) But those issues have been ignored for too long, at tremendous cost, because the serious funding rewards such ignorance. For instance, essentially the same strategy as GWAS is now being pursued on a huge scale, just using a different genotyping technology. No one has run an initial small-scale experiment validating this new approach, it's just based on the hope that the extra information provided by the new technology will provide the ingredient that the studies to date have been missing.