`finaletoolkit agg-bw`¶

1. About¶

Aggregates a bigWig signal over constant-length intervals defined in a BED file.

Every per-base bigWig the pipeline produces — WPS, adjusted WPS, cleavage profile — covers tens of thousands of separate TSS windows. At any single site the signal is sparse and noisy; the structure only becomes visible when the windows are stacked and averaged.

agg-bw does exactly that: it aligns all the intervals on their common coordinate system and averages the signal position by position, producing one profile that represents the sample's average behavior around a TSS. This is the form in which WPS and cleavage results are normally plotted and compared between samples.

2. Where the pipeline uses it¶

The same subcommand runs three times per sample, on three different inputs:

Workflow step	Input bigWig	Output
`agg_wps`	`wps/{sample}_wps_out_tss.bw`	`wps/{sample}_wps_out_tss_aggr.wig`
`agg_adjust_wps`	`adjust_wps/{sample}_wps_out_tss_adjusted.bw`	`adjust_wps/{sample}_wps_out_tss_adj_aggr.wig`
`agg_cleavage_profile`	`cleavage_profile/{sample}_cleavage_profile_tss.bw`	`cleavage_profile/{sample}_cleavage_profile_aggr.wig`

All three are invoked identically apart from input and output paths.

3. As the pipeline runs it¶

finaletoolkit agg-bw <input>.bw <tss_interval> \
    -o <output>.wig \
    --mean \
    -v

3.1 Arguments¶

input_file

The bigWig to aggregate. value: one of the three tracks above

interval_file

The intervals the signal was computed over. value: the tss_interval file for the selected --genome build

These must be constant-length intervals — the aggregation aligns them by position, which is only meaningful if they are all the same size. The bundled tss_interval files are the fixed-width windows around each TSS. See Reference files.

-a, --mean

Average instead of taking the median. set by the pipeline

Passed as the long form --mean. A mean preserves the amplitude of the periodic nucleosome signal linearly, which keeps the aggregate comparable between samples; a median would be more robust to outlier windows but would attenuate the peaks the aggregate exists to reveal. This matches the --mean choice in adjust-wps, so the two stages treat the signal consistently.

The median filter window is not set

agg-bw accepts -m, --median-window-size, whose help suggests setting it to 120 when aggregating WPS signals. The pipeline does not pass it. For the adjusted-WPS track this is intentional — adjust-wps has already applied a 200 bp filter, and a second smoothing pass would over-attenuate the nucleosome peaks. The raw-WPS and cleavage aggregates are therefore unsmoothed, so they retain their high-frequency noise; smooth them at plotting time if needed.

4. Output¶

A wiggle (.wig) file holding the aggregate signal — one value per position across the common interval extent.

These _aggr.wig files are the most immediately usable outputs of the WPS and cleavage analyses. Loaded into a genome browser or plotted directly, the adjusted-WPS aggregate should show the nucleosome-depleted region at the TSS flanked by a regular ~190 bp periodic array.

5. Controlling it¶

What	How
Which intervals	the `tss_interval` entry of `config/genome.json` for the selected build
Aggregation statistic	fixed at `--mean` in the workflow
Threads / memory / walltime	the `agg_wps`, `agg_adjust_wps` and `agg_cleavage_profile` entries of `config/cluster.json`

All three steps share the same modest allocation:

"agg_wps": {
    "threads": 8,
    "mem": "4G",
    "time": "04:00:00",
    "partition": "norm"
}

agg-bw takes no -w, --workers option, so the threads value only sizes the job's CPU request.

6. Requires¶

tss and tss_interval for the aggregation of raw WPS; chrom_sizes additionally for the adjusted-WPS and cleavage aggregates, since their upstream steps need it. Both bundled builds provide all three.

finaletoolkit agg-bw¶