The methods calculates strand cross-correlation profile to determine binding peak separation distance and approximate window size that should be used for binding detection. If quality scores were given for the tags, which quality bins improve the cross-correlation pattern.
get.binding.characteristics(data, srange = c(50, 500), bin = 5,
cluster = NULL, debug = F, min.tag.count = 1000,
acceptance.z.score = 3, remove.tag.anomalies = T,
anomalies.z = 5,accept.all.tags=F)Tag/quality data: output of read.eland.tags or similar function
A range within which the binding peak separation is expected to fall. Should be larger than probe size to avoid artifacts.
Resolution (in basepairs) at which cross-corrrelation should be calculated. bin=1 is ideal, but takes longer to calculate.
optional snow cluster for parallel processing
whether to print debug messages
minimal number of tags on the chromosome to be considered in the cross-correlation calculations
A Z-score used to determine if a given tag quality bin provides significant improvement to the strand cross-correlation
Whether to remove singular tag count peaks prior to calculation. This is recommended, since such positions may distort the cross-correlation profile and increase the necessary computational time.
Z-score for determining if the number of tags at a given position is significantly higher about background, and should be considered an anomaly.
Whether tag alignment quality calculations should be skipped and all available tags should be accepted in the downstream analysis.
Cross-correlation profile as an $x/$y data.frame
Position ($x) and height ($y) of automatically detected cross-correlation peak.
Optimized window half-size for binding detection (based on the width of the cross-correlation peak)
A list structure, describing the effect of inclusion of different tag quality bins on cross-correlation, and a resolution on which bins should be considered.