**alteRNA** is an alternative thermodynamic approach to the *RNA secondary structure prediction problem*,
which aims to minimize a linear combination of total free energy and
total energy density using the dynamic programming formulation proposed
by Alkan et al.(RECOMB 2006), in contrast to the available alternatives
such as Mfold, RNAscf and alifold which all employ the standard
thermodynamic approach.

### Input and Output

alteRNA requires a sequence in FASTA
format where the sequence is represented as a string of characters from
the alphabet {A,C,G,U,T} (the characters are case insensitive). There
are two user specified parameters, sigma and b.
alteRNA outputs the results in three different forms:

- The predicted RNA structure in dot-parenthesis format. The sequence
is given from 5' to 3' end, and the structure is given with matching
parenthesis denoting a base pair and a dot denoting an unbounded base.
- The predicted RNA structure in Connect(.ct) file format. The
sequence length is given in the first line. For each nucleotide i,
there is a line which consists of: the line number(i), the letter
denotion of the nucleotide, the predecessor base index(i-1), the
successor base index(i+1), the paired base index(0 if unpaired) and the
original base index(i).
- The graphics files for the predicted secondary structure in
Postscript(.ps) and GIF format. These graphs are created using NAVIEW
and Mfold/plt2.

### Parameters

#### Sigma Value

Given an RNA secondary structure,

* the energy density of a basepair*
is defined as the free energy of the substructure that starts with the
basepair, normalized by the length of the underlying sequence.

*The energy density* of an unpaired base is then defined to be the

* energy density* of the closest basepair that encloses it.

**alteRNA** thus aims to minimize

*ED(n) + Sigma * E(n)*, where

* ED(n)* is the total energy density of paired and unpaired bases,

*E(n)* is the

* total free energy* and n is the length of the RNA sequence. Here

**Sigma** determines the weight of the contribution of the total free energy in the optimization function. As

*Sigma*
approaches to infinity, the predicted secondary structure gets closer
to that implied by the standard thermodynamic approach (employed by
Mfold).

#### b value

The free energy of a multi-branch loop as implied by
the standard thermodynamic model is not a linear function and as a
result cannot be used in a dynamic programming formulation. Thus

**alteRNA** uses the same approximate formulation as per Mfold, which, for a given multi-branch loop, L sets:

Here a, b, c are constants in thermodynamic model, l_s is the number of
unpaired bases, l_d is the number of base pairs and Sigma Sigma G_stack
is the free energy of each branch in the loop. The

**b value** sets b in this formulation, thus penalizing unpaired bases.