Documentation · ChIP-qPCR
ChIP-qPCR, as % of input
A ChIP batch reports how much of the input chromatin each antibody pulled down at each locus. Each immunoprecipitated fraction is divided by its own input, at the same amplicon and in the same chromatin sample, so there are no reference genes and no calibrator group.
% of input
The published two-step form, which first adjusts the input Cq by log(d) / log(E) and then raises E to ΔCq, is the same formula written in two lines: E^(−log_E d) is exactly 1/d.
The efficiency cancels out of the dilution term and acts on ΔCq alone. At a strongly enriched locus, where the IP comes up close to its input, correcting it changes little. At background, where ΔCq can reach 8 cycles, it changes a lot, and that weak end is where enrichment calls are made. Each amplicon gets the mean LinRegPCR efficiency of its wells, after the same Grubbs test as an expression batch.
Setting up a ChIP batch
Choose ChIP-qPCR as the batch type when you create the batch, then give every well a fraction in the plate layout: input, IP, mock, or NTC. Fractions are recognized from the sample names of the instrument file where possible. The gene field carries the amplicon and the group field the chromatin sample, so a plate with two conditions needs both named. IP and mock wells also take an antibody, and results are grouped by it. The confirmation step lists any IP without an input at the same amplicon and chromatin sample, since those cannot be computed.
The input divisor
d is where ChIP data most often pick up a constant-factor error. A 2% input holds one fiftieth of the chromatin that went into the IP, so d = 50. An input diluted 1:10 gives d = 10. Ampliova asks which of the two you mean and prints the divisor it will apply, because the mistake is invisible in the numbers and rescales every value on the plate.
Replicates and censored IPs
Technical replicates go through the same replicate rule as an expression batch, and the results list the wells it removed. Show the wells under the calculation chain and click one to remove or restore it by hand.
An IP whose replicates never crossed the threshold is reported as an upper bound rather than a zero, computed from the last cycle of the run and marked with a < in the results. When only some replicates amplified, the value rests on those and the analysis tells you to read it as a ceiling.
Error bars default to ± 1 SD of the technical replicates of one ChIP, with SEM, a 95% interval or none as options. All are computed on the Cq scale and mapped back through E, so they come out lopsided, as a ratio does. They describe how precisely the qPCR read one tube. Biological variability comes from independent chromatin preparations, entered as separate chromatin samples.
Not yet in ChIP batches
Statistical tests between conditions, normalization to a control region, PDF and Excel exports, and shared links. ChIP results export as CSV, with the full calculation chain, and as JSON, with every well.
Why no fold enrichment
Ampliova does not compute fold enrichment over an IgG or no-antibody control. Haring and colleagues showed identical ChIP signals producing tenfold differences in that number depending on how much background the control happened to carry, and advised against it [31]. Assign your mock as a fraction and read its % input beside the others instead: the background is then a number of its own rather than a hidden divisor.