As the field has moved from 6-cutter to 4-cutter enzymes, and subsequently dual-enzyme digests, the higher density of sites has made this framework less useful, since it has become increasingly easy to satisfy the intervening site criteria.

The breakdown of separation distance is only calculated for cis-mapping pairs.

Ideally, Hi-C proximity ligation should produce many pairs which are greater than 1000 bp apart. However, these statistics are strongly influenced by the state of the reference. For draft assemblies the distance at which pairs can map is limited by the degree of fragmentation and length of contigs. As a result, many more pairs will be categorised as trans-mapping and pairs which are truly inter-molecular cannot be distinguished from those which are merely inter-contig.

A characteristic of Hi-C libraries, is the presence of a large peak below 1000 bp. qc3C attributes this to regular (and undesirable) shotgun pairs creeping through the Hi-C protocol. The peak is used by qc3C to infer the insert size, which is later employed to estimate unobservable extent of inserts.

Note: the inferred insert size can be significantly smaller than what a sequencing facility might report the experimentally determined insert size to be. This discrepancy can be explained by the failure to account for the additional adapter sequence when fragments are assessed during library preparation.

For trivial single-enzyme digests, there is only one possible junction sequence and so the result for these experiments are not plotted. For dual-enzyme (such as Phase Genomics) there are four potential junctions, while for dual-enzyme digests with one ambiguous site (such as Arima Genomics) there are 16 possible junction sequences.

How efficiently the more complicated library protocols are at producing hybrid junctions is possibly just a point of interest.

Junctions are named for which enzymes was responsible for creating the 5' and 3' ends. E.g. Sau3AI/MluCI would involve two different enzymes, while Sau3AI/Sau3AI only one, as would be the case in a single-enzyme digest. Proceeding the name is the actual junction sequence.

The junctions are grouped by their 5' and then 3' enzyme, while the color spectrum used across each bar aims to emphasise these enzymatic sources.

Note: in BAM mode, the counts are controlled for false positives, in the sense that read alignments must terminate at a cutsite, but the read sequence must continue and contain the observed junction.

Here, Mean adjusted Hi-C fraction represents the best estimate of the proportion of a library's read-pairs which are a product of proximity ligation. This figure is arrived at by correcting the raw estimate for the fraction of insert extent which was not observable.

The observable extent is limited by the length of reads relative to the supplied insert size, as well as a further constraint on flanking sequence around any suspected junction sequence.

For trivial single-enzyme digests, there is only one possible junction sequence and so the 
result for these experiments are not plotted. For dual-enzyme (such as Phase Genomics) there are 
four potential junctions, while for dual-enzyme digests with one ambiguous site (such as Arima 
Genomics) there are 16 possible junction sequences.

How efficiently the more complicated library protocols are at producing hybrid junctions is 
possibly just a point of interest.

Junctions are named for which enzymes was responsible for creating the 5' and 3' ends.
E.g. `Sau3AI/MluCI` would involve two different enzymes, while `Sau3AI/Sau3AI` only one, as 
would be the case in a single-enzyme digest. Proceeding the name is the actual junction 
sequence.

The junctions are grouped by their 5' and then 3' enzyme, while the color spectrum used across 
each bar aims to emphasise these enzymatic sources.

Note: in k-mer mode, the counts are not controlled for false positives.