Averaging measurement strategies for identifying single nucleotide polymorphisms from redundant data sets

Single nucleotide polymorphisms (SNPs) studies have been an active topic of research in the life sciences in recent years. Because SNPs are abundant, stable and sometimes can be related to specific diseases, they have been widely selected as biomarkers for multi-purpose research. As traditional methods for identifying SNPs are time-consuming and expensive, discovering SNPs from expressed sequence tags (ESTs) has became an alternative efficient way. As most EST databases do not store quality/trace files together with EST reads, several methods, like Phard, which requires corresponding sequences quality files, will not be suitable for further research purpose. Thus, computational methods that are able to obtain reliable SNPs without the need for trace/quality information are still essential. We have developed a pipeline framework, called PFSNP, to reveal reliable SNPs from EST data sets without the association of trace/quality files. PFSNP deploys several strategies, like modified neighborhood quality standard measurement and fuzzy logic, in this framework. Also, it automatically adjusts the slide window to efficiently fit different conditions of data sets. PFSNP is demonstrated by identifying SNPs from two subgroups of Oryza sativa with two different strategies as well as zebrafish. Based on our experimental results, PFSNP can obtain higher reliable results when compared to existing methods.