4C profile from an anchor

Identical to its reference in the validationVirtual 4C as cooltools’ virtual4c computes it (Open2C 2024). doi:10.1371/journal.pcbi.1012067

In the app

Analysis panel, under 2D tasks: 4C profile from an anchor

Command

chromosomix fourc

Reads

A contact map

Also known as

virtual 4C, v4C, viewpoint, anchor

A virtual 4C profile is one row of the map: everything one locus, the anchor, touches, as a track along the genome that you can export as a bigWig. It is computed once, over the anchor’s sequence or the whole genome, so panning does not change it.

Running it

Open a contact map, choose 4C profile from an anchor in the 2D tasks card of the Analysis panel, and name the anchor: type a locus, 1-based and inclusive as IGV and samtools write it, such as IV:1,200,001-1,210,000, or a sequence’s name; or press pick on map and Shift-drag along either axis of the map, which fills the box instead of zooming. Leave cis only on to read the anchor’s own sequence, or turn it off for the whole genome. Press run.

The profile is a row of the Tracks card in the Annotations panel, marked * and named after its anchor, 4C · IV:1,200,001-1,210,000, then (with trans), · O/E, · per coverage or · anchor excluded as they apply. Its bigWig, bedGraph and TSV are in the row’s settings. The same anchor run again replaces its row.

The same run from a shell:

chromosomix fourc <map> --anchor <locus> [--<parameter> <value> …] --out <dir>

Parameters

normalisation: by P(s) divides each bin by what its separation from the anchor predicts on the anchor’s own sequence, taking distance out of the shape; by coverage divides it by how much the bin is seen at all, centred on one, so a bin every contact misses is not a dip; by P(s) and coverage does both; none, the default, is the contacts as counted, as in a 4C bigWig. exclude anchor region leaves the anchor and one bin either side empty.

Parameter	Takes	Default	What it does
anchor--anchor <locus>	a locus, display form	None.	—
normalisation--normalise <value>	none, coverage, ps or ps+coverage	none	none — contacts as they were counted; coverage — each bin divided by how much it is seen at all, so a dead bin is not a dip; ps — observed over expected at that separation, from this sequence’s own decay; ps+coverage — both divisions: the decay out of the shape, the depth out of each bin.
cis only--cis / --no-cis	on or off	on	On, this reads one sequence and costs a handful of requests; off, it walks the pixel table for the contacts leaving that sequence.
exclude anchor region--excludeAnchor / --no-excludeAnchor	on or off	off	Leave the anchor and one bin either side empty, so its contact with itself does not set the scale.
use balancing weights--no-balance	a switch	On wherever the file carries balancing weights; --no-balance reads the raw counts.	—

What it writes

• profile: a value per bin: TSV, bedGraph and bigWig

On the command line every file goes into --out as <file stem>.fourc.<output>.<ext>, and each TSV opens with # lines naming the build, the file and every parameter that produced it.

Reading the result

Raw, a bin’s value is its contacts with the anchor, summed over the anchor’s bins, balanced where the file has weights. On a log scale it is a triangle with its apex at the anchor, and the contacts you are after are bumps on its flanks.

Over P(s), the profile is centred on 1: a bin above 1 touches the anchor more than its distance predicts. Off the anchor’s sequence, the expected is the mean of what the anchor reaches on the other sequences.

Dips left on a raw map are mostly coverage, such as repeats and bins with few restriction sites, exactly where a spurious contact is most tempting. by coverage takes them out, most usefully on a raw map, since balancing does much the same.

The anchor’s own bins are the brightest by orders of magnitude and set an autoscaled track’s scale; exclude anchor region leaves them out.

Gaps and zeros. With cis only on, every bin off the anchor’s sequence is a gap, not 0. A bin with no stored contact with the anchor is 0, and a bin the balancer dropped is a gap.

Compared with cooltools

In the validation, over the whole genome, the profile is cooltools.virtual4c’s to float32, except at the viewpoint itself and where cooltools leaves a bin without a value: chromosomix writes 0 there.