This is a test version of Biostars. For the public version, visit https://www.biostars.org.
In Peak Calling, what is the input sample

Every time I've done a ChIP-Seq, or (as in this case) CLIP-Seq, analysis over the years, I have used the "input" samples and not used the "control" samples. Whenever I've had both, I would google the difference to see which I should use and everything always said to use the input in whichever case I've encountered. However, now I only have control samples and I don't have input samples and I'm not sure how to call peaks on them.

First of all, dumb question: what is the input sample? Is it the same as the sample that is passed through a column for enrichment via antibody, but it simply wasn't passed through the column? Or is it the sample run through a column that has only beads and no antibody?

The controls I have, as they were described to me, were run through a column that had beads and no antibody. I was going to use PEAKachu to call peaks, but I'm not sure what I should do differently with these control samples. Can anyone clear this up for me? I've been putting off trying to figure out the difference between input & control for years now.

clip-seq chip-seq
• 1,980 views • link
updated by colin.kern colin.kern • written by rleach rleach

1 answer

From my understanding, a control ChIP-seq library is an umbrella term for all the libraries you're describing, and also including using a control antibody such as IgG. An input library is a specific type of control library that's essentially just making a library from a portion shearing/sonication output without using a control antibody. I'm actually not sure if these are generally put through a column with beads but no antibody, or just directly made into a library. I'm more on the computational side of things.

But regardless, in your peak calling analysis these should all be treated basically the same. However the control and/or input was created, it's supposed to act as a way to normalize your actual ChIP library against various biases that come from shearing, mappability, etc. So even if the software specifically asks for an input rather than the control, I would still provide whatever control you've generated even if it technically isn't an input library.

• 0 views • link
written by colin.kern colin.kern

That's reassuring. Thanks. Is there a way to assess how good a control/input is? I basically want to figure out whether I should suggest that they change their procedure to sequence the input as their control.

• 0 views • link
written by rleach rleach

Log in to answer this question.

More posts like this