Lecture 3
Comparative analysis of RNA secondary structure
	The undisputed "Gold Standard" in secondary structure determination (NOT prediction!)
	Principle
		comparison to genetic analysis of mutation/secondary reversion
	Early examples
		DNA - Chargaf's rules & Watson/Crick
		tRNA - cloverleaf only common structure possible for first three tRNA seqs
	Analysis of secondary structure - an interative process
		essentially a manual process
		refine alignment and add sequences to alignment
			alignments
				definition - 2D matrix - homology
				seq gaps
				basepair designators
				group by phylogeny
			start with only seqs that can be aligned on the basis of sequence similarity
		simple identification of covariations
			use these to add more disparate sequences (or sequence regions)
			large gaps as structural subdomains
			variable regions w/ gaps - often hairpins with variable lengths
		analysis of subsets 
		example - early P seqs in gamma-proteos & Bacillus
		genetic analysis for "confirmation"
			Haas Science paper example
	Statistical analysis of covariations - Mxy
		How to quantitate sequence variation?
			e.g. 50% conservation - but what about the rest?
			information theory - Hx definition
			sequence logos
				P RNA example
		How to quantitate sequence covariation?
			Mxy definition/algorithm
			3D/2D plots of secondary structure
			1D plots of adjacent bps
			basic example - tRNA
		Natural pops approach
			secondary structure details - archaeal P RNA
			non-Watson-Crick pairings - archaeal P RNA
		interpreting data phylogenetically
	Automated methods
		require most of the info sought!, i.e.: 
			starting alignment, or
			seqs than can be aligned based solely on similarity, or
			a starting structure
		the alignment contains the structure as implicit information!
	Data requirements & resolution
		more sequences, better resolution
		need at least ~30 sequences - 50 better, thousands best
	Strengths
		objective, quantitative
		automatable & visualizable
		basepair resolution
		can distinguish thermodynamically equivalent possibilities
		only biologically-relevant structures identified
	Weaknesses
		phylogenetic affects - but can be dealt with
		seq sample affects - i.e. P w/ half purples
		alignment basically a manual process
		Mxy best for final stages of secondary analysis & analysis of tertiaries
		identifies base-base interactions
		no specific information from invariant sequences
		no specific information from idiosyncratic sequences - use MFOLD!
		difficult to incorporate biochemical data