Sleuth analysis with only using Kallisto abundance.tsv files no HDF5 files

Question

[weixiaokuan]

Hi, I have a set of results (abundance.tsv) from Kallisto and want to use sleuth to analyze them.
All these results are in one directory and per @dsfull suggestion, I created a dataframe with path col indication where these files are. This solution can get the analysis moving forward but I got new error.

"Error in H5Fopen(file, flags = flags, fapl = fapl, native = native) : HDF5. File accessibility. Unable to open file."

Is this because I don't have HDF5 file for each abundance tsv? Could I just use abundance.tsv to run sleuth without HDF5?

Thank you. -Xiaokuan

[dsull]

You'd specify -DUSE_HDF5=ON in cmake when compiling kallisto from source (though that option is usually by default enabled in the kallisto binary releases). You then run kallisto quant and specify the bootstrap option via -b. That will get you hdf5 files.

However, if you want to use tsv files, make sure that every file name ends in .tsv

Otherwise, show me your sleuth_prep command and the entire output (not just the error message), and I'll try to figure out what's going on.

For kallisto abundance.tsv (aka no bootstraps), I'd recommend deseq2 or limma-voom for differential gene expression. sleuth really only has the advantage if you do bootstrapping.