I think the macs2 tutorial is a little vague because there are many options for "control" sample.
For example, in my lab we are using input DNA as the control.
By that, I mean: DNA that you treated exactly the same way as the IP samples.
ChIP pipeline is:
- Cross-link protein to DNA (formaldehyde)
- Sonicate DNA (chop into fragments)
- Immunoprecipitate w/ antibody against transcription factor of interest
- Make libraries, sequence.
To get input, sequence the output from step #2. It controls for uneven sonication and cross-linking.
However, you can get more creative with this.
For example, let's say you have a line that expresses a mutant allele of your transcription factor - maybe it lacks a phosophorylation site you think is important for activation. Or maybe it has a mutatation that makes the phosphorylation site appear constitutively active. (A phosphomimic.)
You could put this line through steps 1 to 4 and then use the resulting sequences as the control in a macs2 analysis.
A cautionary note: When working with biology collaborators to design such experiments, make sure it will work for them by analyzing data sets from their species that have a similar design. Or be extra conservative and tell them to sequence both types of control.
The control is the input DNA. You can use a mock sample (IgG) but the recommendations from ENCODE and others is to use input, as it represents the true background and can help correct for bias.