PsiCLASS is a reference-based transcriptome assembler for single or multiple RNA-seq samples. Unlike conventional methods that analyze each sample separately and then merge the outcomes to create a unified set of meta-annotations, PsiCLASS takes a multi-sample approach, simultaneously analyzing all RNA-seq data sets in an experiment. PsiCLASS is both a transcript assembler and a meta-assembler, producing separate transcript sets for the individual samples and a unified set of meta-annotations. The algorithmic underpinnings of PsiCLASS include using a global subexon splice graph, statistical cross-sample feature (intron, subexon) selection methods, and an efficient dynamic programming algorithm to select a subset of transcripts from among those encoded in the graph, based on the read support in each sample. Lastly, the set of meta-annotations is selected from among the transcripts generated for individual samples by voting. While PsiCLASS is highly accurate and efficient for medium-to-large collections of RNA-seq data, its accuracy is equally high for small RNA-seq data sets (2-10 samples) and is competitive to reference methods for single samples. Additionally, its performance is robust with the aggregation method used, including the built-in voting and assembly-based approaches such as StringTie-merge and TACO. Therefore, it can be effectively used as a multi-sample and as a single-sample assembler, as well as in conventional assemble-and-merge protocols.
module spider psiclass to find out what environment modules are available for this application.
- HPC_PSICLASS_DIR - installation directory
- HPC_PSICLASS_BIN - executable directory
- HPC_PSICLASS_EXE - example directory
If you publish research that uses psiclass you have to cite it as follows: