Experimental Plans

What is an experimental plan?

An experimental plan is a description of your samples, what you plan to measure about them, and what are the controls and experimental variables you are using.

We structure an experimental plan as follows.  First there is an investigation which will hold the whole plan.  Connected to the investigation are one or more studies.  Each study contains one or more conditions and one or more assays. 

A condition is a summary of the process of going from the organism or cell line to the extract you use in performing an assay.  It includes treatments, tissue, organ, genetic background, time of day, age etc.

An assay is what technique you are going to apply to the sample including possible variations.  In a typical ChIP-seq experiment, you will have an assay for the ChIP libraries, and another for the input library.  If you ChIP for multiple targets, then each target will have its own assay.

A study is a combination of conditions and assays that are typically focussed on a particular part or stage of the experiment.  The website presents a study as a table of conditions versus assays.  Usually all of the cells in the table will be filled in by the end of the experiment.  This gives a nice visual representation of the progress of the experiment.

Multiple studies can be put together under one investigation as needed.

Why do I need to make one?

For a few reasons.  

First, although we have long collected information such as organism, tissue, developmental stage, condition, we sometimes get samples where this information does not capture important aspects of the sample that are important to the experiment.  So, we needed to make our sample description capture more information.

Second, the Core is frequently called upon to submit data to public repositories as part of getting your publication accepted.  Having these experimental plans, and gathering detailed data up front will help make the publication process easier.

Third, as sequencing capacity has increased experiments have become more complex.  Also, with time, investigators have accumulated a large number of samples, and these need a more coherent system to organize them and the subsequent analyses. We will use the experimental plans as a way to help organize, find, and share data.

Is it hard to make a plan?

No, its very easy.

To make a plan submit a description to the Core using either the on-line form or email.  We will write up a 'formal' plan then confer with you to finalize the plan.

Our goal is to capture the technical details of the experiment, but to still present the results using the terms or jargon that you normally use when discussing the experiment.  For example, if you are knocking out  Myod in heart muscle at E14.5 we can capture that technical detail,  but give that condition a tag or nickname of 'KO'.  Similarly, what you call the 'WT' in the experiment, is probably not a wild mouse gathered from a field in the Poconos, but rather may be a BL6 with a loxP flanked gene.  We will record this, but still give such a mouse the nickname 'WT' (or what ever you normally call it.)

What about my old experiments?

We have used the PI, investigator, and sample information we have to estimate the design of all existing experiments.  These estimated designs have been loaded into the database and are being used in the new website.

We know that the estimated designs are flawed and ugly.  Also you may have several separate designs that really belong together as one.

Please feel free to setup a time to meet to correct earlier experimental plans - use the appointment calendar link to find a time we can meet.

What are Source IDs for?

The purpose of the source ID is to allow us to know which libraries come from the same biological source. The source ID can include

• anonymized donor IDs
  
• your local mouse IDs,
  
• sample prep IDs
  

Tracking source IDs will allow us to distinguish biological from technical replicates, help guide analyses.

Here are some examples of what we have in mind.

In an experiment on WT and KO mouse livers, source IDs such as WT1, WT2, WT3, WT4 KO1, KO2, KO3, KO4 are fine. They let us know that each sample came from a different mouse.

If you are testing a library or sample prep method on the same sample, and we have included the testing in the experiment design, you should use the same ID for all of the samples.