Nestler lab epigenomics data share

Investigator-administered cocaine mice

ChIP-seq NAc 7d24h

H3K4me3

RNA-seq NAc 7d24h

Self-administered cocaine mice

ChIP-seq NAc 14d24h

H3K4me3

Social defeated mice

ChIP-seq NAc 10d48h

H3K4me3

H4K16ac

NOTE: Differential sites were predicted by a window p-value <1E-4 and were not further filtered. You are encouraged to filter them by FDR(column padj) and fold change(column logFC) using Excel or similar program before use. Commonly used value for padj are 0.1(10%) or 0.05(5%) and it does NOT hurt to increase to 0.15 or 0.2 if you are getting too few targets.

The "Treatment.avg" and "Control.avg" columns: Sometimes a differential site may contain very large or even infinite fold change but very low abundance. They are likely background noise and should be removed. These two columns serve this purpose and they are basically the average of "Treatment.cnt" and "Control.cnt".

How to choose cutoff for read abundance: Usually you can draw a histogram for the averaged count and set a point where everything below seems to be much smaller than the rest. If you do not feel like to do this, a ballpark estimate of 30 or sth. similar can be used.

About TDF: TDF(.tdf) is a binary format read by the IGV genome browser. It contains the genome-wide coverage information for an NGS sample(such as ChIP-seq and RNA-seq), which can be displayed as curves, histograms or heatmaps in a genome browser.

BED file is a tab-delimited text format which contains at least three columns. It can also be read and displayed by IGV. A differential list can be easily convered to a BED by selecting the first three columns and removing the header line(s)(i.e., keep only the genomic coordinates).

The IGV genome browser can be downloaded here. It is a Java program which runs on all operating systems.

Tips for using IGV:

• There are many options for customizing visulization under menu: view -> preferences and in right mouse menu.
• Multiple tracks can be selected at the same time in the left track panel using the Ctrl or Shift key.
• To normalize the coverage(Y-axis), check "Normalize Coverage Data" in "Tracks" tab. This makes Y values comparable across multiple samples(because they have unequal total read numbers). You may need to reload your data after making the change.
• Usually you want to uncheck the "Autoscale" option(in right mouse menu) because it makes background noise look like binding enrichment.
• To make multiple tracks comparable, use the same values in "Set Data Range"(in right mouse menu) for all tracks.
• Each differential list has been annotated using RefSeq or Ensembl gene database. RefSeq is more stable than Ensembl so it is more commonly used here.