Notes from eMeeting on Nov. 30, 3007
1. Review Randy's proposed thoughts about storyline
- Consider which might be implemented as this project, and which held for future proposals
- Consider in light of grant proposal goals and achievability
2. Outline initial programming steps and parameters for Stephen and co.
3. Decide on rough initial development schedule and when, generally, to meet in person
Paul H. demo of mouse Biologica setup
Mouse pictures - Tub gene is dominant for obese mice
Set up genes under chromosomes
TUB different from tub - C changes to a T to be only difference in two genes
Pedigree - can cross and make families of 3-5 offspring. Can cross and look at individual chromosomes of mice.
Easy to view multiple generations and parents
Showing how to create cpecies:
"My World": record of all species created in past is tracked,
Chromosomes included here and properties
Don't have mutations or markers/introns, etc.
Genetics is determined by rules: If -
CD: Polygenic possible -- how to QTL -- Could we do weighted linear combinations?
Randy -- Huge potential here -- don't name them "genes", but call them "markers" This will allow us to have multipple -- THis allows us to have multiple strains and look at the markers.
ALso can do haplotyping -- done with snips within individual pieces of DNA -- somewhere in the middle of 26-base-pair region (snip) is one thing that is different. Look for snips outside of genes -- pepper this with snips to ididentify the strains. Allows us to make markers and snips to get through QTL analysis most of the way.
Paul: Software doesn't care if piece is an encoding region or not. Want region of snip to be close enough to the gene to be useful but not interfere with the coding region. Be careful not to lead students to think that snips are coding for the phenotypes
Can also do recombination in meiosis
Randy -- two other things:
1) If a snip is inside of a gene, it may not dstroy the function of a gene (more than one codon set per protein) Does, however, create ghost QTL that swamps out the real effects. Therefore, we exclude snips inside genes to avoid this.
Also, numbering genes is done from long to short for historical reasons.
Randy -- if you were doing a QTL analysis within this tool, you would have a library of strains and markers -- look for two strains in this library -- pick strain that has high blood press/salt and one that doesn't. Look in library and find 40 markers in library different between strains. Generate 400 animals, meausre blood pressure, measure genotype at each marker position. ENd up with 200-300 markers with genotype info and equal no. of phenotype info. This goes in a matrix that gets passed to QTL statistics program for searching for statistical correlations. This model is completely represented in this system piecewise.
Paul: THe making of that matrix isn't there -- we can make markers, (need to distinguish b/w markers and genes) Rules more complicated. Might need to fake a few things, but not many obvious problems here.
Randy: First pass -- stick with Mendelian trait for year one and avoid complex trait analysis? Introducing to HS students -- good place to start. Also, analyzing complex trait loci is complicated, and scientists don't have it down...
Stephen: recap -- 3-week activity
week 1: dragons and drakes - Mendelian
week2: mice -- also Mendelian?
week3: actually explore something we didn't know -- have class decide where to focus research in this week?
Does this require more tools, or are we okay with this toolset?
Randy -- dragons/drakes good to get involved -- strong storyline necessary. Combining this with fundamental principles of science important. Teaching fundamental skills of science. Closeup magic -- thimbles. Elegant and beautiful -- what is the point is important?
Gross outline of possible storyline important -- more time with dragons? Return to do this in session
Chad -- make sure we get to data at some point --
Stephen -- not necessarily get to end of this whole thing -- have this end up as open, but not necessarily make open-ended piece as complete as originally sketched out?
Randy -- look at record of breeding initially -- just like scientific model. Constraints same as before -- more realistic model of real-world science. When we get into organizing breeding scheme -- how much do they pay and what do they choose - haplotype mapping and others. Dont' expose things that they wouldn't have ready acdess to with mouse data. AFter 2-3 week program with dragons and drakes, give them access to pools of mouse data and then let them explore. May be a team that gets to next stage, which is more focused group?
Stephen -- pieces of software can change things -- may not expse these at all. Trying to match the dragons and mice is important. May find that the students need background knowledge elements
Randy -- haplotype mapping, knockouts, others 6 core principles in a few hours... ?! In a mouse, 7 million snips, 35,000 genes, etc. Complicated system par excellence. Dragons and drakes give us a model system more controlled
Stephen -- when kids do just a little bit, they often don't get things - bite off a few important ideas and use them. Find the important pieces for QTL. If not, then go with the elements that are.
Randy: elemts we need -- QTL -- include link to phenotype here
Haplotype is the background and genetic identity piece here
knockouts are a post-QTL process
QTL with mice here -- can do dragons Mendelian-wise and introduce complex traits with mice
Stephen: When we go to mice, students can comprehend that some things are the same, and that some things are much, much more complex, adn to imagine that it gets even more involved with humans -- scale much larger in humans.
Randy: software won't work with mice, probably. Will integrate with genome browser, will model QTL, etc.
Going to mouse stage -- envision givein them ten data sets and ten publications exploring diseases in mice and show them how QTLs are done. Can give them genome browser and ask them to show region of chromosome 4 related to heart disease. Read description of research and look into their relations. Take principles of huge human-mouse dataset adn relate them back to the pieces
Stephen: What differences between software and approaches, for general information?
Randy: I don't see model as sufficeintly complex, and we don't know enough about mouse genetics to tell you how to model it within this system. A while back, they found a marker that tells you where to start and stop reading in genes. We don't know what half of these do, and what they make, etc. WE also don't know how they get turned on and off. Coding is 3% and other 97% is unknown ("junk"?!). All happening fast -- Science/Nature paper compared 6 fly strains and found 150 genes that didn't use standard stop point. Even more genes that only did some of the pieces. Some genes made 5 different proteins...
Stephen: Could imagine adding functionality that looked like genome browser, etc. But can't model or dump mouse strains into the program/software. Could add capabilities to make it as direct an analogue to the real world elements. Then focus on using external sites and get a sense of complexity of using real models. Scaffold these investigations to real websites for scie3ntists.
Randy:; Where leveraged easioly -- GBrowse -- take interface and use with dragons and drakes If introducing to tools we are using with mice -- already know technique. Difference in magnitude only? Initial script
topok ideas from science and mapped to universe. Need ot think of mouse script first. First introduce principles, then allow then to reproduce others' research. Seeing how data became publication leads them to see how they can do this in the third step.
Stephen -- our next steps -- get basic technologies up. Paul pulled out of the archive and made it work. Take work Randy has done -- Chad synthesize goals and pieces to move us forward. When do we integrate this to a system that can be best for the project but also mesh with Concord pieces. Other -- probably can't do within project, but: Brown Bag lunch Tim Erickson from Epistemological Engineering -- WEb application could probably see with a note to him -- Group -- we were scientists investigating a universe of a 12x12 grid of squares. Could do experiments on a 3x3 grid and spend money and publish papers and write conjectures; submit and accepted or rejected by the person. Could see data for others -- (Mastermind/Battleship 2.0? --CD) Also can discuss them -- collaborative aspects that would be incredible in the long run.
Randy : Big rocks to pick early on -- Start to write some short descriptions for key items such as haplotype analysis, QTL analysis, knockouts. First draft from Randy. Also: dragons and drakes will need to generate table ready to go into QTL analysis (includes markers, intercross with 200-400 dragons with known genotype and phenotype)
Stephen: Meet on this with Randy and Paul and possibly p;rogrammer
RAndy -- will write up a description of a typical
Next steps:Meeting --
10:00-11:30 on Wednesday, Dec. 12
Randy will post notes about the needs for a QTL matrix of data by Dec. 9th
Chad will begin a very rough draft of initial instructional steps and inquiry scaffolding.
Stephen will look into the next software steps and go over the QTL data needs with Paul H. before the meeting on Dec. 12th.
Stephen and Chad will look at GBrowse - Stephen to think about the possibilities for integration and Chad about the interface presentation to students in the instructional process.