"Design of Lightweight Stochastic Context Free Grammars
for RNA Secondary Structure Prediction"
-------------------------------------------------------
(Notes summary on how CONUS was utilized for the paper)
=======================================================

Assumptions:
------------
CONUS code has been compiled and executables are in 
a directory called "bin/". 

The three datasets are in a directory called "data/"
and have the names:
 1) mixed80.stk		LSU/SSU dataset for training
 2) rfam5.stk		Rfam v5 seed info for ambiguity testing
 3) benchmark.stk	Benchmark dataset (RNase P, SRP, tmRNA)

Grammar Correspondance:
-----------------------
The grammar codes CONUS uses are as follows:
CONUS grammar code	Paper's Grammar
	NUS			G1
	YRN			G2
	UNA			G3
	RUN			G4
	IVO			G5
	BJK			G6
	UYN			G7
	RY3			G8
	BK2			G6S

Training:
---------
bin/conus_train -g NUS -s nusmixed80.mod data/mixed80.stk

This produces a binary output file called nusmixed80.mod which
contains the NUS model trained on the mixed80.stk dataset.

Ambiguity Experiments:
----------------------
bin/reorder -m nusmixed80.mod -b N data/all_rfam5.stk

where N is the number of suboptimals to generate and test.
Must have a trained model file for input.  Only available 
on the two ambiguous grammars (NUS and YRN).

Ambiguity Testing:
------------------
bin/ambtest -i -m unamixed80.mod data/all_rfam5.stk

Will halt as soon as any sequence/structure pair fails the
ambiguity test.  Note that the -i option asks ambtest to 
ignore the input structure and generate a test structure by
CYK.  

Plus One Grammar Check:
-----------------------
bin/pocheck -m unamixed80.mod data/all_rfam5.stk

This verifies that all grammars give appropriately
equivalent score under a simple +1 scoring scheme.
[Note that the additional restriction of not permitting 
lone pairs means that UYN and RY3 may not equal the 
other 7 grammars; but they should equal each other.]

Plus One Grammar Performance:
-----------------------------

bin/conus_fold -q -w 2 -g NUS data/benchmark.stk > nus.stk
sreformat --wussify stockholm nus.out > nus_w.stk
compstruct data/benchmark.stk nus_w.stk

bin/conus_compare -w 2 -g NUS data/benchmark.stk

Trained Grammar Performance:
----------------------------

bin/conus_fold -q -w 2 -m nusmixed80.mod data/benchmark.stk > nus.stk
sreformat --wussify stockholm nus.out > nus_w.stk
compstruct data/benchmark.stk nus_w.stk

bin/conus_compare -w 2 -m nusmixed80.mod data/benchmark.stk