Region list
| Document kind | region |
| Drawn from | a file — it can also be previewed in the file manager |
| Formats that open here | BED, AED |
| Refuses to draw below | 240px wide |
What it draws
A list of named intervals on a genome, as a document in its own right.
A region list is what a laboratory keeps when it has decided where to look: a panel of genes, a set of recurrent cytogenetic regions, a browser track of calls exported from somewhere else. It is not a sequence, and it is not a result. It says that these coordinates are interesting, and usually who says so and why.
The canvas draws it two ways, and the control switches between them without leaving the tab:
- the karyogram — every region marked in the gutter beside the chromosome it sits on, on all twenty-four chromosomes or only the ones the list touches;
- the table — every region with its locus, the bands it covers, its length, and every other column the file states.
Above both, the canvas states what the file says about itself: the track's own name, how many regions it holds, how many chromosomes they fall on, how much sequence they cover in total, how many data lines the file had, and the assembly — or that it states none.
What opens in it
BED — the UCSC browser's tab-separated interval format. Its three required fields are the sequence name, the start and the end; everything past those is optional, and what a file states is carried rather than dropped.
AED — Affymetrix Extensible Data, the other format a cytogenetics workstation writes for the same purpose. AED exists so that a laboratory can add its own typed columns beside each region, and those columns appear in the table under the names the file gives them, never renamed.
A file that holds no readable data lines is refused with the reason, not drawn as an empty list. Lines that could not be read as regions are listed with what was wrong with each, and when there are more than twenty the canvas says how many it is not showing.
What this computes
Very little, deliberately: a region list is a statement, and this canvas is where you read it. Intersecting a list against a sample's own findings is a different question and belongs to the copy-number tool, which reads the same file through the same reader.
What is derived here:
- Total span — the sum of every region's length. It is the sum of the intervals as stated, and overlapping regions are counted once each, so a list whose regions overlap covers less unique sequence than this number. It is a property of the list, not a measurement of a genome.
- Bands — which of the assembly's cytogenetic bands each region covers, read off the band table for the assembly in force. A region inside one band reads that band; one crossing several reads its first and its last. This is the plain fact underneath a cytogenetic designation, not the designation itself: naming a finding in ISCN is a separate, standardised act, and the canvas does not perform it.
- Chromosomes the list touches, and which sequence names the assembly in force does not carry. Regions on an unrecognised sequence stay in the table and are not drawn, and the canvas says how many there are rather than dropping them.
The assembly, and why the canvas keeps asking about it
A region list does not carry its own genome. BED's required columns are the sequence name, the start and the end, and none of them is the assembly; a genome browser supplies it from whatever session you opened the file in. The same coordinates mean different places on GRCh37 and GRCh38 — megabases apart on most chromosomes — so a build assumed silently is a whole class of quiet error.
So the canvas distinguishes five states, and says which one you are in every time:
| The state | What the canvas says |
|---|---|
| The file states a build we have a band table for | that build is in force and it is the file's claim, not yours |
| You chose one for a file that states none | that build is in force and it is your assertion |
| You turned bands off | what the file states is repeated, and nothing is drawn on an assembly |
| The file states a build we ship no band table for | it is named, and nothing is chosen for you — choosing another would assert an assembly the file contradicts |
| Nobody has stated one | the coordinates are exact, and there are no bands, no chromosome lengths and no band names |
In every one of them the coordinates are exact, because those are what the file actually holds. What an assembly buys is the picture: banded chromosomes, true relative lengths, and the band each region falls in.
A BED can state a build, on its optional track line. An AED states it as a named property. Both
are read when present.
Uncertainty and quality
- A region list carries no dosage score, no pathogenicity and no result. Where a file states none, the table shows none — it is not filled in from elsewhere. A curated list that does carry scores shows them in its own columns under its own names.
- No region is marked as found or missed on the karyogram, because there is no sample here to find it in. The marks say someone chose to look here, and they carry no status colour for that reason.
- A curated designation and the bands an interval covers can disagree, and when they do you can see both: the region's own name is one column and the derived bands are the next. Across the ClinGen dosage-sensitivity panel these agree exactly for 42 of 51 regions on GRCh37 and 44 of 51 on GRCh38; the rest are places where the curators named a region at a coarser resolution than its coordinates occupy, and one where the published interval and the published label are different places on both builds.
Controls
- View — karyogram or table. The karyogram is disabled with its reason when no assembly is in force, rather than offered and then empty.
- Assembly — the build the regions are drawn on. Its default is whatever the file states; where the file states nothing, nothing is chosen for you.
- Only chromosomes with regions — off draws all twenty-four, so a chromosome this list covers nothing on is visible as such rather than absent.
- Chromosome size — how tall the chromosomes are drawn. It changes nothing the file states.
- Find region — keeps the rows whose name, identifier or chromosome contains what you type, and says how many of how many matched.
Every one of these is available in the file-manager preview as well as in the full tab.
Where it appears
- As a stage tab, opened by clicking the file in the project's file manager.
- As a file-manager preview, drawn by the same component with the same controls, read-only.
The same file is also an input elsewhere: the copy-number tool reads a region list against a sample's own calls, to report what that sample has in each region. It is read there by the same reader that draws it here, so the two cannot disagree about what the file says.
Performance and limits
- Below 240px wide the canvas refuses to draw. The table needs room for the region, its locus, its bands and its length side by side before a row means anything.
- The karyogram draws twelve chromosomes per row, which is how a karyogram is read, and six in a narrow pane.
- Nothing here writes to the file. A region list opened as a document is read-only; editing a panel is done where the panel is kept.
Screenshots
No composed figure of this canvas exists yet. The pictures on this site are made for the page they sit on rather than captured from a session.
Related
- The copy-number canvas is where a region list becomes a result: the same regions, intersected against one sample's own calls.
- The sequence map is where a file that carries bases opens. A region list carries none, which is the distinction this canvas exists to keep.