Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 | 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 6x 6x 6x 6x 6x 6x 6x 6x 6x 6x 5x 5x 2x 4x 4x 4x 4x 4x 4x 1x 1x 1x 3x 3x 3x 3x 3x 6x 6x 6x 6x 1x 1x 1x 6x 6x 6x 6x 6x 6x 6x 3x 3x 3x 3x 4x 4x 4x 4x 4x 4x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 6x 6x 6x 6x 6x 3x 6x 6x 6x 6x 6x 6x 6x 3x 3x 3x 3x 1x 1x 1x 1x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 1x 1x 1x 1x 4x 4x 4x 4x 4x 4x 4x 4x 4x 4x 4x 4x 4x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 4x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x 3x | /**
* The origin and terminus, run — in the annotations worker, beside the
* residues and the coverage track.
*
* Per sequence: the GC skew in windows of `window` bases every `step`, its
* cumulative sum and that sum's two turning points; the parS and matS sites;
* the coverage sampled at 5 kb and smoothed by a cubic spline, where a coverage
* track was chosen; and then `callOriTer`, which is the whole of the decision
* and is tested on its own (`analysis/oriter.ts`).
*
* The sequences are the map's where there is one, so the loci land on the map's
* names; with no map they are the reference's own. A declared-circular sequence
* (§8.15) is windowed round its origin and its sites clustered as angles, which
* matters exactly here: most assemblies start at `dnaA`, so the origin is the
* one locus a linear reading would split in two.
*/
import { FeatureBuilder } from '../../annotations/featureset'
import { compileMotif, findMotif, isMotifRefusal, type Motif } from '../../annotations/motif'
import {
baseCounts,
cumulativeOf,
gcSkewOf,
windowCounts,
} from '../../annotations/composition'
import {
callOriTer,
oriTerRatio,
skewTurns,
splineFit,
type OriTerCall,
type TerminusRule,
} from '../oriter'
import { idParam, numParam, tickedSequences, type TrackTaskRunner } from '../job'
import { sitePattern } from '../tasks/oriter'
import type {
IntervalsOutput,
ProfileOutput,
TableOutput,
TaskOutput,
} from '../types'
import { gridContigs } from './motif'
/** The resolution coverage is sampled at before it is smoothed, as Bacchus does. */
const COVERAGE_STEP = 5000
/** How the table says which step decided the terminus. */
const RULE_WORDS: Record<TerminusRule, string> = {
matS: 'nearest the matS cluster',
parS: 'farthest from the parS cluster',
coverage: 'lower coverage',
skew: 'skew maximum',
}
/** Sites found on one sequence: midpoints for the call, and the intervals to draw. */
interface Sites {
mids: number[]
starts: number[]
ends: number[]
strands: number[]
mismatches: number[]
}
function sitesOf(seq: Uint8Array, motif: Motif | null, mismatches: number): Sites {
const out: Sites = { mids: [], starts: [], ends: [], strands: [], mismatches: [] }
if (motif === null) return out
// Both strands: a binding site is where it is whichever strand the assembly
// wrote it on, and both consensuses are palindromes, reported once per place.
for (const hit of findMotif(seq, motif, { bothStrands: true, mismatches }).hits) {
out.mids.push((hit.start + hit.end) / 2)
out.starts.push(hit.start)
out.ends.push(hit.end)
out.strands.push(hit.strand)
out.mismatches.push(hit.mismatches)
}
return out
}
export const runOriTer: TrackTaskRunner = async (job) => {
const entry = idParam(job.params, 'sequence')
if (job.residues === undefined) {
job.note('This session cannot read residues, so there is no reference to read.')
return []
}
if (entry === null) {
job.note('No reference was chosen. Load the genome’s FASTA and pick it under Reference.')
return []
}
const reader = job.residues(entry)
if (reader === undefined) {
job.note('The reference chosen is no longer loaded. Drop the FASTA again, or pick another.')
return []
}
if (reader.refusal !== undefined) {
job.note(reader.refusal)
return []
}
// The two motifs, either of which may be left empty to skip its step.
const motifs: Record<'parS' | 'matS', { motif: Motif | null; pattern: string; mismatches: number }> = {
parS: { motif: null, pattern: '', mismatches: 0 },
matS: { motif: null, pattern: '', mismatches: 0 },
}
for (const key of ['parS', 'matS'] as const) {
const pattern = sitePattern(job.params, key)
if (pattern === '') continue
const compiled = compileMotif(pattern)
if (isMotifRefusal(compiled)) {
job.note(`${key}: ${compiled.reason} Its step was skipped.`)
continue
}
const asked = Math.round(numParam(job.params, `${key}Mismatches`, key === 'parS' ? 1 : 0))
motifs[key] = {
motif: compiled,
pattern,
mismatches: Math.max(0, Math.min(compiled.length - 1, asked)),
}
}
const window = Math.max(100, Math.round(numParam(job.params, 'window', 5000)))
const step = Math.max(10, Math.min(window, Math.round(numParam(job.params, 'step', 1000))))
const knots = Math.max(4, Math.min(40, Math.round(numParam(job.params, 'knots', 10))))
const coverageEntry = idParam(job.params, 'coverage')
const circular = new Set(job.framing.circular ?? [])
// The map's sequences where there is a map, under the map's names; the
// reference's own otherwise. Ticked narrows either.
const grid = gridContigs(job)
const everyName = grid.length > 0 ? grid.map((contig) => contig.name) : reader.contigs.map((c) => c.name)
const ticked = tickedSequences(job.params, 'sequences')
const names = ticked === null ? everyName : everyName.filter((name) => ticked.includes(name))
if (names.length === 0) {
job.note('No sequence is ticked, so there is nothing to read.')
return []
}
const loci = new FeatureBuilder()
const parSites = new FeatureBuilder()
const matSites = new FeatureBuilder()
let parCount = 0
let matCount = 0
const rows: Array<Record<string, string | number>> = []
const fits = new Map<string, (at: number) => number>()
const skipped: string[] = []
for (let i = 0; i < names.length; i++) {
const name = names[i]
job.progress(i, names.length, name)
const seq = reader.bases(name)
if (seq === undefined) {
skipped.push(name)
continue
}
const length = seq.length
// A skew needs a few windows to have a shape at all.
if (length < 4 * window) {
skipped.push(name)
continue
}
const round = circular.has(name)
const counts = baseCounts(seq, step)
const skew = gcSkewOf(windowCounts(counts, Math.max(1, Math.round(window / step)), round))
const turns = skewTurns(cumulativeOf(skew), step, length)
if (!Number.isFinite(turns.low) || !Number.isFinite(turns.high)) {
skipped.push(name)
continue
}
const par = sitesOf(seq, motifs.parS.motif, motifs.parS.mismatches)
const mat = sitesOf(seq, motifs.matS.motif, motifs.matS.mismatches)
let coverageAt: ((at: number) => number) | undefined
let ratio = Number.NaN
if (coverageEntry !== null) {
const sampled = await job.sample(
coverageEntry,
[{ contig: name, start: 0, end: length }],
COVERAGE_STEP,
)
const region = sampled.regions[0]
if (region !== undefined && region.values.length > 3) {
const positions = new Float64Array(region.values.length)
const values = new Float64Array(region.values.length)
for (let k = 0; k < region.values.length; k++) {
positions[k] = (region.starts[k] + region.ends[k]) / 2
values[k] = region.values[k]
}
const fit = splineFit(positions, values, length, knots, round)
coverageAt = fit.at
fits.set(name, fit.at)
ratio = oriTerRatio(fit.fitted).ratio
}
}
const call: OriTerCall = callOriTer({
length,
circular: round,
low: turns.low,
high: turns.high,
parS: par.mids,
matS: mat.mids,
...(coverageAt === undefined ? {} : { coverageAt }),
})
// A locus a window wide, centred on where the call put it: the skew places
// a turning point to within a window, and a one-base interval would claim a
// precision the method has not got.
const half = Math.floor(window / 2)
const detail = `${RULE_WORDS[call.rule]}${call.agreesWithSkew ? '' : ' · disagrees with the skew'}`
for (const [label, at] of [
['ori', call.ori],
['ter', call.ter],
] as const) {
loci.add(name, Math.max(0, at - half), Math.min(length, at + half), {
name: label,
detail: label === 'ter' ? `terminus · ${detail}` : 'origin',
})
}
for (let k = 0; k < par.starts.length; k++) {
parSites.add(name, par.starts[k], par.ends[k], {
strand: par.strands[k],
name: 'parS',
score: par.mismatches[k],
detail: `${par.mismatches[k]} ${par.mismatches[k] === 1 ? 'mismatch' : 'mismatches'}`,
})
}
for (let k = 0; k < mat.starts.length; k++) {
matSites.add(name, mat.starts[k], mat.ends[k], {
strand: mat.strands[k],
name: 'matS',
score: mat.mismatches[k],
detail: `${mat.mismatches[k]} ${mat.mismatches[k] === 1 ? 'mismatch' : 'mismatches'}`,
})
}
parCount += par.starts.length
matCount += mat.starts.length
if (!call.agreesWithSkew) {
job.note(
`${name}: the terminus was placed ${RULE_WORDS[call.rule]}, at the cumulative skew’s ` +
'minimum — which is where the skew alone would put the origin. Check the assembly, or ' +
'the motifs.',
)
}
rows.push({
sequence: name,
length,
topology: round ? 'circular' : 'linear',
// Display form: 1-based, as every coordinate a reader sees (§26.1).
ori: call.ori + 1,
ter: call.ter + 1,
rule: RULE_WORDS[call.rule],
parS: par.starts.length,
parSCluster: call.parS?.count ?? 0,
matS: mat.starts.length,
ratio,
})
}
job.progress(names.length, names.length, 'done')
if (skipped.length > 0) {
job.note(
`${skipped.length.toLocaleString()} ${skipped.length === 1 ? 'sequence was' : 'sequences were'} ` +
`not placed — not in the reference, or shorter than four skew windows: ` +
`${skipped.slice(0, 6).join(', ')}${skipped.length > 6 ? ', …' : ''}.`,
)
}
if (rows.length === 0) return []
if (coverageEntry === null) {
job.note('No coverage track was chosen, so there is no ori/ter ratio and no coverage step.')
}
const linear = rows.filter((row) => row.topology === 'linear').length
if (linear > 0) {
job.note(
`${linear.toLocaleString()} ${linear === 1 ? 'sequence was' : 'sequences were'} read as ` +
'linear. A bacterial chromosome is usually circular — tick it in the Chromosomes table and ' +
'run this again, or the origin at its first base is split between its two ends.',
)
}
const reference = job.inputs.labels?.[entry] ?? 'reference'
const outputs: TaskOutput[] = [
{
kind: 'intervals',
id: 'oriter-loci',
label: `Origin and terminus · ${reference}`,
payload: { kind: 'features', data: loci.finish() },
noun: { one: 'locus', many: 'loci' },
// Places, as the sites below are: a track beside the map, and a point on
// it where the reader puts them there (§18.7).
display: 'lane',
look: 'point',
} satisfies IntervalsOutput,
]
if (parCount > 0) {
outputs.push({
kind: 'intervals',
id: 'oriter-pars',
label: `parS sites · ${motifs.parS.pattern}`,
payload: { kind: 'features', data: parSites.finish() },
noun: { one: 'site', many: 'sites' },
display: 'lane',
look: 'point',
})
}
if (matCount > 0) {
outputs.push({
kind: 'intervals',
id: 'oriter-mats',
label: `matS sites · ${motifs.matS.pattern}`,
payload: { kind: 'features', data: matSites.finish() },
noun: { one: 'site', many: 'sites' },
display: 'lane',
look: 'point',
})
}
// The smoothed coverage on the map's own bins, where there is a grid to put
// it on — the curve the ratio was read from, to be looked at beside the raw.
const binsize = job.framing.binsize
if (fits.size > 0 && binsize > 0 && grid.length > 0) {
const total = grid.reduce((sum, contig) => sum + Math.max(1, Math.ceil(contig.length / binsize)), 0)
const values = new Float32Array(total).fill(Number.NaN)
let startBin = 0
for (const contig of grid) {
const bins = Math.max(1, Math.ceil(contig.length / binsize))
const at = fits.get(contig.name)
if (at !== undefined) {
for (let b = 0; b < bins; b++) {
values[startBin + b] = at(Math.min(contig.length - 1, (b + 0.5) * binsize))
}
}
startBin += bins
}
const coverageLabel = job.inputs.labels?.[coverageEntry ?? ''] ?? 'coverage'
outputs.push({
kind: 'profile',
id: 'oriter-coverage-fit',
label: `${coverageLabel} · spline fit`,
values,
extent: [0, total],
binsize,
unit: '',
} satisfies ProfileOutput)
}
outputs.push({
kind: 'table',
id: 'oriter-table',
label: 'Origin and terminus per sequence',
columns: [
{ key: 'sequence', label: 'sequence', kind: 'text' },
{ key: 'length', label: 'length', kind: 'number' },
{ key: 'topology', label: 'read as', kind: 'text' },
{ key: 'ori', label: 'ori', kind: 'number', hint: 'Centre of the origin locus, 1-based.' },
{ key: 'ter', label: 'ter', kind: 'number', hint: 'Centre of the terminus locus, 1-based.' },
{ key: 'rule', label: 'terminus placed by', kind: 'text' },
{ key: 'parS', label: 'parS', kind: 'number', hint: 'Sites found on this sequence.' },
{
key: 'parSCluster',
label: 'in cluster',
kind: 'number',
hint: 'parS sites within an eighth of the sequence of the densest one.',
},
{ key: 'matS', label: 'matS', kind: 'number', hint: 'Sites found on this sequence.' },
{
key: 'ratio',
label: 'ori/ter',
kind: 'number',
digits: 2,
hint: 'Highest over lowest smoothed coverage: copies of the origin per copy of the terminus.',
},
],
rows,
} satisfies TableOutput)
return outputs
}
|