The use of JavaScript to develop an implementation of the alignment-free [ 5] map-reduce decomposition of sequence analysis could be justified solely on grounds of convenience.
Applied Biosystems (personal communication) confirmed that this reference bias exists in the current implementation of the Bioscope alignment algorithm, and efforts are underway to mitigate this isue in future implementations.
It is a classical implementation of the progressive alignment method.
While 6 total reads is well below the number necessary for determining statistically significant AEI at any one SNP [ 30], at this point we are testing the implementation of the IUPAC alignment across the entire dataset for obvious bias.
It is a bit paradoxical since its implementation of the progressive alignment significantly differs from the canonical one, in that it delays the incorporation of the most distantly related sequences until the second and unique iteration.
Gapped alignments were performed using a constrained, fail-fast, semiglobal implementation of the dynamic programming alignment algorithm (the Needleman-Wunsch algorithm (Needleman and Wunsch 1970), modified to not penalize initial gaps and to trace back from the highest-scoring terminal node of either sequence).
Briefly, a set of 3 812 663 proteins from RefSeq version 21 consisting of all predicted proteins encoded in 458 completely sequenced and unfinished genomes and 254 609 proteins from Swiss-Prot version 51.5 were compared, in a pairwise manner, with the program SSEARCH (http://fasta.bioch.virginia.edu/), an implementation of the SW local alignment algorithm.
A C-implementation of the local alignment kernel was provided by Saigo et al. [58].
During implementation of the SW algorithm, the alignment sequences must be stored in the global memory and then moved to the local memory of a multiprocessor.
The implementation of such alignment processes can constitute a significant task.
