I want to do the differential expression analysis of RNA-seq data. My sample does not contain biological replicates. I tried using partek, but it does not calculate differential expression at gene level without biological replicates.
You can use DESeq, a module in R software. It is done to use with biological replicates, but you can also use it without replicates (it is explained how to do it in the DESeq doc). You don't need to be an expert in R to use it, and it is as the moment one of the most popular statistical analysis for RNAseq data. You can find it easily on bioconductor on the web.
Consider to use the edgeR, a package of R/Bioconductor and one of popular DE analysis tools for RNA-Seq data. It has a good documentation with several study cases as example. Mailing list of edgeR is a good place for you to look at either. You can have a look on the differential expression analysis for non biological replicate RNA-Seq data in the user guides.
I have used DESeq which is a part of R package and can be downloaded through bioconductor! I heard good things about CLC Genomics Workbench that can be used to analyze RNA-Seq data from start to finish using visual interface, however it's expensive! Good luck!
The most common solution is probably the Tophat/Cufflinks package. The included Cuffdiff program does differential expression. Easy to use. http://cufflinks.cbcb.umd.edu/
I think Cufflinks is good but it can become a little bit confusing with all the isoforms when all you really want is to measure gene expreesion , in the sense of transcription rate. I tried both DEseq and Cufflinks. DEseq is just simpler and gave us good validation
About CLC Genomics, i use it also for mapping and RNA seq analysis. Unfortunately you can only work with PRKM and not assess stiatistical variance across replicates, an approach that you have in DESeq and EdgeR that gives you a better assessment of data variation. Another thing i point out in CLC, the mapping excludes a lot of sequences (25%) due to a "too high" segregation of errors, while unix softwares as ssaha2 that uses k-mer approaches seems to integer more sequences so you end up with better statistics. But I might have not look too deep into CLC to modify mapping parameters. About EdgeR, did somebody compared DESeq and EdgeR and can give a feedback on results using the 2 methods?
This one shows very nicely, and I agree with them that variance calculated from n numbers of Bootstrapping would be really helpful to calculate degs more accurately
GFold and DEGSeq should be possible options for single replicate and for cuffdiff the option poisson dispersion should work. For DESeq sort of R packages you can use method=blind in dispersion calculation.