Improving the Mapping of Smith-Waterman Sequence Database Searches onto CUDA-Enabled GPUs

Pseudocode 1

The pseudocode of the scoring function in our proposed method.

/ASSUMPTION

Query sequence length →  , aligned to 4 bytes and padded with dummy residues

Subject sequence length → , aligned to 4 bytes and padded with dummy residues

/

for (; ; += 4)

 Initialize all the relevant variables;

 Load the packed 4 residues between and from texture memory to register;

 for (; ;   += 4)

  Get the to residues of the subject sequence from register;

  Load substitution scores for cells (, ) to (, ) from query profile;

  for (; ; ++)

   Load and values of the cell (, ) from shared memory;

   Compute the , and values of the cells from (, ) to (, ),

   and calculate the maximum score;

   Save and values of the cell (, ) to shared memory;