Sequence variation and stress responses

Via evolgen, a look at the question of whether cis (i.e. noncoding, regulatory sequence changes) or trans (i.e. coding sequence changes) play a greater role in the divergence of species. (Lucky that I stumbled across this blog post this morning because my P.I. brought it up when I went to talk with her this afternoon.) The weight seems to currently lie with people who argue that cis-regulatory variation is less likely to be deleterious, since trans mutations often have pleiotropic effects (i.e. affect multiple phenotypic traits).

Lemos et al., 2008: So the key point about the argument advanced in this paper seems to be that trans effects are more likely to be masked in a heterozygous background in Mendelian fashion, while cis effects tend to be additive. Hence, positive selection can act more effectively on cis-regulatory mutations when they are at very low allele frequencies.

We argue that cis- and trans-regulation undergo distinct population genetic dynamics across short and long timescales, which lead to a relative overabundance of trans-regulation within population and a relative overabundance of cis-regulation between populations. This inference follows from two observations. First, the mutation variance for trans-variation is substantially larger than the mutation variance for cis-variation. [...]

Second, our findings that differences due to trans-regulation show higher degrees of dominance, whereas cis-variation arises from regulatory loci that are more additive (or with rare dominant alleles) may suggest a simple way by which too much trans within populations and too much cis between populations can be reconciled. Accordingly, despite selection against trans-regulatory variation within species being particularly strong because of a presumably larger pleiotropic effect of these mutations (13), substantial recessive variation with large trans-effects might still be maintained concealed in heterozygous in natural populations under mutation-selection balance. However, although cis-regulatory variation is produced at a slower rate than trans-variation, positive selection may act most efficiently on cis-regulatory variation, because allelic variation underlying cis differences might have greater additivity such that differences because of cis loci are less sensitive to genomic background.

The authors basically looked at chromosomal substitution lines (which are only genetically variant at one chromosome and hence are good for large-scale mapping, but that’s neither here nor there) and their F1 progeny, but it would be interesting to see if this model holds true in natural populations. E.g. in yeast, we have full genome sequences of several yeast species and many yeast strains within species. It’s not difficult to identify coding polymorphisms when you have well-annotated genomes; it’s slightly harder to identify potential cis-regulatory sequence variation. What’s most difficult though is the problem of whether those sequence differences are functionally relevant and have undergone selection. Naturally, you can use sequence analysis tools to estimate selection but once again, it boils down to the problem of connecting genotype to phenotype…

Other papers I read/skimmed today:

Kvitek et al., 2008: The authors phenotyped 52 S. cerevisiae strains (mostly natural isolates) under different environmental stresses. They also measured gene expression for some of their strains and found that the s288c lab strain has a markedly different transcription profile (not news: P.I.’s postdoctoral work was able to map most of these differences to major mutations, such as the Ty insertion in HAP1). However, they were also able to find major expression differences within the set of natural isolates as well, which drives home the only real point of the paper: there’s a wide range of phenotypic variation among the available S. cerevisiae strains. Although they were able to demonstrate that some of this variation can be caused by copy number (by comparing strains with duplicated or missing chromosomes), they didn’t succeed in actually doing an association mapping study, probably because not all the strains used were sequenced. I wonder why they didn’t try hybridizing to a SNP genotyping array though? They did do “an associative study to identify gene expression patterns correlated with environmental sensitivity across the 17 non-laboratory strains,” but that unfortunately doesn’t say anything about the sequence changes causing the phenotypic variation (whether it be at the level of transcription or at the level of growth under stress) in the first place. Still, it’s good to have confirmation that transcription does correspond to visible growth phenotypes.

Gasch, 2007: A review by the last author of the paper above looking at an identifiable “environmental stress response” that is shared among the major yeasts, i.e. S. cerevisiae, S. pombe and C. albicans.

Shalem et al., 2008: Paper looking at mRNA production and decay rates under a transient (as opposed to a chronic) stress response. I need to think more about this paper; I think I’m going to propose it for discussion in seminar. The methods section is also incredibly useful since I’m also interested in assaying dynamics for my own project.

Yeast comparative genomics

Cliften et al., 2003 and Kellis et al., 2003 seem to be the major papers for the sequenced Saccharomyces species (S. bayanus, S. kudriavzevii, S. mikatae and S. paradoxus, plus the more divergent S. castelli and S. kluyverii).

The actual data though are on the accompanying websites at Washington University and Broad Institute.

The Sanger Institute also recently sequenced 37 strains of S. cerevisiae and 27 strains of S. paradoxus. Their project page is here. (I still need to email the person in charge of this project about getting access to some of their sequence analysis statistics.)

In the reading queue

Shalem et al., 2008: Comparison of transcriptional profiles under fast and slow stress responses.

Thiebaut et al., 2008: Production of unstable transcripts regulated by nucleotide supply.

Doniger et al., 2008: Neutral and deleterious polymorphisms in two natural isolates of yeast. (Paper doesn’t look well-written, but the methods section may be useful. Also, two more sequenced strains available for natural variation studies.)

Babbitt and Kim, 2008: Test for selection on noncoding sequence, in this case, specifically nucleosome-binding sites. (Sequence-based models of nucleosome binding are still controversial at best, but paper may still be useful.)

Crombach and Hogewag, 2008: Evolution of evolvability in gene network simulations. (Looks interesting! Should I propose this paper for my systems biology seminar?)

Cai et al., 2008: Gene regulation via frequency of bursts of transcription factor localization to nucleus. (P.I. recommended reading this paper.)

R functions

Our lab uses R for all statistical analysis. I’ve gotten fairly adept at writing R scripts, but there are a few functions (or parameters of functions) that I tend to constantly look up. I’m listing them here for my own reference.

your.table <- read.table(file, header, skip, nrow, na.strings, row.names, col.names, ...)
The skip and nrow parameters help you control which part of the file gets read. The na.strings ensures that null entries (however they are denoted in your data file) are read as NA. The last two, row.names and col.names, should be self-explanatory. Many other parameters too, but these are the ones that I use the most often.

ls(all=T)
Lists all objects in the workspace.

rm(ls(all=T))
Clears the workspace.

save.image("filename.rda")
Saves the workspace to a file. I need to use this function more often.

par(new, mfrow, ...)
The new parameter determines whether you open a new graphics window or plot over the preceding plot. (Axes of the succeeding plot should match the former for the graphs to coincide neatly.) The mfrow parameter is specified by a vector c(nrow, ncol) that specifies how many plots to organize in a window and in what configuration.

wilcox.test(...)$p.value
t.test(...)$p.value
Returns just the p-value calculated by the Wilcoxon rank-sum or student’s t test.

Freezer malfunction

Last night, I went into lab to streak out some strains from the glycerol stocks we keep in the -80° C freezer. When I went to return the plate, I noticed that the temperature hadn’t dropped back to its setpoint. I got a little worried and went to consult someone in the neighboring lab about it. She said that the freezer itself looked fine, and I should probably leave it for a while to give it time to stabilize.

I went back to my bench and set up some reactions. When I returned to the freezer to check on it, the temperature had gone up to -69° C. At this point, I was starting to get very concerned and consulted with the neighboring lab again. One post-doc made calls on my behalf to her own (very senior) P.I. at his home to see if he knew of any emergency freezers in the building. A senior grad student from another neighboring lab took a look at the freezer and said that he thought nothing seemed obviously wrong but that I might want to move the racks out just to be safe. He suggested that the reason for the malfunction was that our freezer had so much empty space (it was only about half-full), which made it difficult to maintain a steady temperature. I managed to get in touch with my own P.I. at her home. She agreed that I should move out the racks and promised to come to lab herself to help troubleshoot the problem with the freezer.

Having never faced a situation like that before, I was very grateful to have our helpful neighbors take charge and organize the freezer “evacuation”. They located a freezer with some empty shelf space upstairs, helped me move all the racks, and packed them with dry ice to help the second freezer quickly stabilize. My P.I. decided to fill our malfunctioning freezer with containers of water in hopes that minimizing the empty space would solve the problem. I didn’t manage to get much more done that night after dealing with the emergency. The whole situation was rather unnerving, since my initial reaction was to worry whether I had somehow broken the freezer, and then after we relocated all our racks, I wondered whether I was making a mountain out of a molehill and the freezer would have been fine. Of course, these problems only ever happen late at night or on weekends (or both, as in my case), when we have no way of contacting actual maintenance personnel who know how to actually diagnose these problems. But my P.I. came by this afternoon to let me know that the freezer was now at +8° C, so I had been right in being paranoid. Though I’m concerned that our freezer is indeed seriously malfunctioning, I’m also kind of relieved that all of last night’s hullabaloo wasn’t for nothing.

Lessons learned:

• Be thankful for neighboring labs filled with workaholics who work late nights on weekends and are willing to help you out with problems.
  
• Be thankful for a P.I. who lives nearby and is willing to come in and help you deal with emergencies.
  
• Be thankful that other labs will be sympathetic and not mind if you have to suddenly use their freezer space when no one is around to give permission.
  
• For future emergencies, one should have home/cell phone numbers for everyone in the lab.

It was a harrowing experience, but it did give me enough of an adrenaline rush that I stopped being depressed about my failed experiments (or rather, I am still depressed but feel more energized to keep working), and it also made me think that people are really kind and decent after all. The fact that everyone was willing to stop their own work and help me out really made me have faith in this system. People often tell me that academia and basic science research is highly politicized and secretly competitive; I don’t doubt it, but I also think that nonetheless, people do fundamentally try to help one another out. It’s a reassuring thought, in any case.

As a tangent, my P.I. asked if I often went home from lab so late at night, and I had trouble admitting that I did. It’s a grad student stereotype to pull crazy stunts by working late in lab, and we will often indirectly boast when we loudly complain of the long hours we spend at the bench. But I’ve grown reluctant to admit the full extent of my usual working hours. Grad students who spend excessive hours in lab fall into two categories: the “workaholics” who progress quickly by getting many experiments done and the “procrastinators” who work extremely inefficiently and take twelve hours to do what other students finish in eight. To be honest, I fall in the latter category, so I’m more embarrassed than proud when I end up going home from lab at, say, midnight. Occasionally, I’ve spent the night in lab and gone home at 5 AM before anyone comes in so I can change and pretend that I went home the night before. It’s especially embarrassing when I feel that I have so little to show for my efforts; my labmates, who usually leave at more reasonable hours, seem to get much more done than me.

I hope that by the end of the summer, I’ll have learned to work more efficiently and have more to show for my efforts.

Objectives

I initially created this blog to write about my research interests but I haven’t gotten into the habit of updating it regularly. Part of the reason is sheer lack of time (research grows more and more time-consuming by the week, although I have made little actual progress) and distraction.

But my second year of graduate school is beginning in a few weeks, and I wanted to sit down and make a more serious commitment to updating this blog. I’m going to be a graduate student instructor (GSI) for a course in genetics in the fall, and I suspect that might provide more material. I wanted to outline, however, some personal academic goals, both science-related and not.

For research

• Revive journal club
  
• Update this blog once a week with links to interesting papers (and hopefully some discussion)
  
• Read Uri Alon’s An Introduction to Systems Biology: Design Principles of Biological Circuits
  
• Read Michael Lynch’s Genetics and Analysis of Quantitative Traits
  
• Self-study biostatistics
  
• Master the R-qtl package

For school

• Apply to NSF, Hertz and Soros
  
• Take teaching seriously
  
• Decide on topic for outside proposal (by September)

For self

• Learn Python and Unix bash shell programming
  
• Learn 漢字
  
• Practice translating Korean poetry/short stories
  
• Read all unread purchased books
  
• Keep up book club

An ambitious list of resolutions always means that I will fail to keep more than half of them, but I hope that writing it down in a post will help me stay on track.

Mentorship styles

Via Bitesize Bio, an interesting perspective on Postdoc life vs PI life. Of course, it’s possibly a little premature to be thinking about such issues when I’ve just finished my first year in graduate school, but the first paragraph in particular caught my eye:

The postdoc (and grad student, for that matter) perspective is — and I recognize it both from my own experience and my colleagues — that you come up with most of the ideas, do all the work, and the PI just takes the credit. It is a bit exaggerated, but the point is that the vast majority of the time spend on a project is time spent by a postdoc or grad student. The advisor reads the occasional draft paper, comment on the research a bit here and there, but doesn’t really put in the hours.

A few weeks ago, I chose my thesis lab and jumped right back into benchwork, picking up my rotation project from where I left off at the end of last November. My P.I. conceived the project, tailoring it to my specific interests (there’s a reason why no other roton worked on the same project), and most of the ideas have emerged as a collaboration. In any case, I don’t feel as if my P.I. “just takes the credit”: though I obviously do all the actual experimental work, she provides me with input at every level starting from suggestions about adjusting protocols to the larger picture of what this project aims to contribute to the field. She’s very careful to acknowledge my intellectual input when talking about the project to third parties, and I feel that I have sufficient independence to take the project in directions that interest me. So far, I’ve felt that I’m receiving the right level of guidance while also being encouraged to take initiative and explore ideas on my own.

I wonder if my satisfaction with her mentorship style will decline over the years. Disillusionment with graduate school (cf. Grad School Malaise) seems to be inevitable; what makes the difference is whether you regard it as a temporary phase induced by stress and sense of failure or whether you feel that you made the wrong decision to pursue a career in research. That being said, I do benefit from the fact that my P.I. is a fairly new faculty member, whose lab is small enough that she has enough time to spend guiding each student. I also think that compatibility of personalities helps.

My other two rotations had P.I.s with extremely different mentorship styles: the hands-off P.I. who only touched base with his students’ projects if they had an urgent question or during their lab meeting and the bullying but attention-deficit P.I. who goes around arguing with his students over what experiments they should do next (and berating them for not working long enough hours). Ironically, I think both P.I.s can actually be ideal for certain students, depending on what sort of mentorship they’re looking for. Personally, I think I could have managed in both labs but I would not have been as happy; one of the reasons why I chose this particular lab over both was because I thought the P.I. struck the right balance in her mentorship style.

C. elegans genomics

Blumenthal et al., 2002: Genomic analysis of operons by microarrays. Authors probed for SL2 sequence, which is recognized by a snRNP responsible for trans-splicing of the polycistronic mRNA. They used SL2/polyA ratio to determine percentage of genes downstream in operons. Their positive control set: confirmed SL2 genes, negative control set: genes first in operon. They assessed distribution of operons in the genome, the number of genes per operon, and distribution of intercistronic distances within operons.

Why have operons? Is it more efficient? (Some operons in yeast, I believe, but not sure whether at a similar frequency as in C. elegans.) Does it relate to synteny at all? Operons ususally consist of genes with similar function or similar regulation, so there must be an evolutionary benefit. Then why not have operons? Authors suggest that it may have to do with the compactness of the C. elegans genome. (How compact is compact? More compact than, say, Drosophila, which doesn’t have operons?) May be present in genomes of other nematode species, which is interesting. Must come hand-in-hand with trans-splicing machinery in eukaryotes.

Kamath et al., 2003: Construction of RNAi library for C. elegans, covering 86% of predicted genes in genome. Authors kept track of knockdown phenotypes, organized into phenotypic classes, which show enrichment for certain gene functions. How stringent is the knockdown? How consistent is the effect on phenotype? Interesting to see the chromosomal distribution of the phenotypic classes, especially in light of the operons mentioned above.