Quick usage

Complete mode

If you're working with raw data (IDAT files), you can run phyfum in complete mode. It will preprocess the files with minfi. If needed and if both tumor and normal samples are available, it will also run a copy number analysis with rascal. This will allow to blacklist fCpGs that won't fluctuate as the model expects.

An example run for this workflow would look like this:

phyfum run --input ${input_dir}/epic_array_dir \
    --output experiment1 \
    --workdir experiment1 \
    --patientinfo /path/to/epic_array_dir/sample_sheet.csv \
    --patient-col patient \
    --age-col age \
    --patient-col Patient \ 
    --sample-col Sample \
    --sample-type-col group \
    --stemcells 3-10-3 

Trees mode

In case you already have the beta values, you can run phyfum in trees mode. The pipeline will simply deploy the XMLcreator tool to format the input data as expected by BEAST and run the inference.

phyfum run --input /path/to/exampleBeta.csv \
    --output onlybetas \
    --workdir onlybetas \
    --patientinfo /path/to/meta.csv \
    --patient-col patient \
    --age-col age \
    --sample-type-col group \
    --stemcells 3-10-3

Phyfum auto-detects what kind of input is provided and selects automatically the optimal workflow.

A custom set of fCpG sites can be provided. Otherwise, an additional step to select the fluctuating sites will be ran (at least 2 individuals are required).

CpGs fluctuate differently across tissues. Therefore, we recommend not reusing fCpG sets accross different tissues