Scalogram

Not yet compared with its referenceThe scalogram of Lioy et al. 2018. doi:10.1016/j.cell.2017.12.027

In the app

Analysis panel, under 2D tasks: Scalogram

Command

chromosomix scalogram

Reads

A contact map

Also known as

reach, distance, quantiles

The scalogram shows how far each bin’s contacts reach: at every bin, the separation within which a tenth, a quarter, a half, three quarters or nine tenths of its contacts lie, drawn along the genome as nested bands, as Lioy et al. (2018) and HiContacts’ scalogram() draw it. Where P(s) averages a sequence into one curve, the scalogram shows which stretch folds differently from the rest.

Running it

Open a contact map, choose Scalogram in the 2D tasks card of the Analysis panel, tick the sequences, choose the quantiles and press run. The bands arrive on the Tracks card of the Annotations panel, a row each — Scalogram 10 %, Scalogram 25 % and so on — grouped as Scalogram in one lane, on a logarithmic axis in base pairs.

The same run from a shell:

chromosomix scalogram <map> [--<parameter> <value> …] --out <dir>

Parameters

look out to, at 0, reads each sequence whole: quick on a bacterial or a yeast map, a large read on a mammalian chromosome, and the only setting at which a band is a quantile of all of a bin’s contacts. Any other reach, in bins, makes each band a quantile of what lies within it, and its name says so: Scalogram 50 % · within 500 kb. diagonals to ignore leaves out each bin’s contact with itself, which would pull every band towards zero; minimum coverage of the distribution leaves a bin without bands where too few of its separations have a value.

Parameter	Takes	Default	What it does
General
quantiles--quantiles <value>	standard, quartiles or deciles	standard	One band per quantile, nested in a single lane. standard — lioy et al.’s bands, and what HiContacts draws; quartiles — the quartiles alone; deciles — a finer picture; nine lanes of ink.
look out to--reach <n>	a whole number, at least 0, in bins	0	0 is the whole sequence; a cut makes every quantile a quantile of what is within it, and the lane says so.
sequences--sequences <name,…>	sequence names, comma-separated	Every sequence of the file when left out.	—
use balancing weights--no-balance	a switch	On wherever the file carries balancing weights; --no-balance reads the raw counts.	—
Quality
diagonals to ignore--ignoreDiags <n>	a whole number, at least 0	1	1 drops the bin’s contact with itself, which is not a contact at a distance.
minimum coverage of the distribution--minValid <n>	a number, 0 to 1	0.5	Below this fraction of usable separations the bin gets no bands at all.

What it writes

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

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

Reading the result

Each band is a separation: at the 50 % band, half of everything the bin touches lies closer than the band’s height. A low, narrow bundle is a stretch whose contacts are local; a high one, a stretch whose contacts reach far; a step in every band at one place, a boundary between two ways of folding. The median does not move when the balancer drops one distant cell, where the mean can move by a factor of two, so a deep map and a shallow one compare.

A hole in the bands is a bin with too little of its distribution to measure, such as beside a stretch the balancer dropped; a band exists only where the one below it does too.

What to watch for

• The order of the rows is the picture. Each band fills down to the one below it in the group: reorder them and the picture changes.
• Circular sequences. A separation is the shorter way round, at most half the sequence.
• Short sequences, of fewer than four bins, are left out, with a note.
• The values are base pairs. Only the lane is logarithmic; the bigWig, bedGraph and TSV hold the separations as computed.