David Laehnemann

I asked over at `CoinUtils` -- seeing that they handle the `COIN-OR` file format handling, they should know which formats are widely used and well-specified: https://github.com/coin-or/CoinUtils/discussions/149

One of the CoinUtils developers gave his perspective on MILP file formats: https://github.com/coin-or/CoinUtils/discussions/149#discussioncomment-401745 From it, I take it that he would agree with you, in that the `MPS` format is...

What's the entry in `config["calling"]["scenario"]` (which is usually read in from `config/config.yaml`) and what does the file specified by this entry contain? From what I can see from the traceback,...

All the adaptations you might have to do should be to the files in the `config/` folder: https://github.com/snakemake-workflows/rna-seq-kallisto-sleuth#step-2-configure-workflow In `config/config.yaml` you can set your reference at: https://github.com/snakemake-workflows/rna-seq-kallisto-sleuth/blob/3cd43dbaaea80c53a952807c0869a65de6679479/config/config.yaml#L6 And you can...

Sure, let's discuss further. My first instinct would be for technical replicates to be modelled via the batch variables...

Yes, generally you can use this workflow to call variants on any type of sample setup, including the case where you only have a tumor sample. You can even specify...

For the candidate callers, all you should care about is sensitivity / recall, not precision, as varlociraptor will take care of that and you can easily filter results in a...

Nice, thanks a lot for investigating and documenting this here! For freebayes, are you using the parameters as they are specified in this workflow? https://github.com/snakemake-workflows/dna-seq-varlociraptor/blob/52a139f04bc03dbb9c05db3dbfaa723f1f603092/workflow/rules/candidate_calling.smk#L14-L17 Namely, this turns off genotyping...

The short answer is `yes`. ;) But maybe this quick example helps for a good intuition of how the FDR control works: The posterior probabilities that varlociraptor provides for all...

The general FDR approach basically sorts `1 - PROB_EVENT` from the smallest to the largest, and finds a cutoff so that the sum of all the sites up to that...