Rules for Path Traversal of Contigs in an Isogroup

5.2.2

5.2.2.1

Contigs lacking reads connecting them to other contigs on one or both ends are used to initiate the traversal of an individual isotig. When a contig has no reads connecting it to any other contigs, it may become an isotig composed of a single contig (see Figure 104).

If the –isplit option is used, isotigs will also be initiated when a depth “spike” of at least 50% between one alignment column and the next is found. When a “spike” in depth is found while traversing an isotig path through the alignment for a contig (see Section 5.2.2.2), the alignment column in which the relative depth change occurs is marked for future use as an initiation point for further isotig traversal. The isotig path up to that point is reported as an isotig, and the original isotig’s path continues until it terminates. The marked initiation point is then used as a starting contig and all paths starting from it are reported until they are exhausted.

Figure 104: Isotig initiation

5.2.2.2

Spike Detection

Identification of a spike along a path occurs when all of the following conditions are met at the end of a contig where a spike is detected:

•	The alignment depth is at least 10 reads

•	A minimum of 20% of the aligned reads must be in the opposite orientation relative to the more abundant orientation of the aligned reads

•	A spike may not occur within 10 bases of an already detected spike

•	A change in alignment depth between one alignment column and the next of at least 50% signals the location of a “spike”.

Figure 105: Spike identification

5.2.2.3

Path Extension

An isotig path that starts at an initiating contig (call this Contig I) will be extended by following reads that consistently lead to another contig. Once a terminating condition is found on a particular path, the isotig is reported, and then all other paths starting from Contig I are reported until they are exhausted. The process used to explore all possible paths is called “recursion.” Extension along a particular path will continue until a terminating condition is met (see Section 5.2.2.4).

5.2.2.4

Path Termination

The end of an isotig is found when any of the following conditions occur:

•	No reads are found that extend from the contig currently at the end of the isotig path

•	The number of reads connecting two contigs is less than 5% of the alignment depth of either

•	The Isotig Contig Count Threshold is reached. In this case, the further traversal of a particular isotig in an isogroup will be stopped.

•

A contig is reached whose length is below the Isotig Contig Length Threshold. If a contig is reached with a length shorter than this threshold, the further traversal of a particular isotig in an isogroup will be stopped. The contig shorter than the icl threshold will be marked as such and reported in the output files.

•

A cyclic path is encountered. Recursive path traversal will stop if cyclic structures are detected, i.e. revisiting one contig which has already been included in an earlier part of the isotig being traversed. Such cyclic structures will be marked in the output files by assigning cyclic status for the first contig detected.

5.2.2.5

Other Rules that Prevent Path Traversal

•	If the number of contigs in an isogroup exceeds the Isogroup Threshold, the paths in the isogroup will not be traversed to report isotigs.

•	When the number of isotigs in an isogroup is found to exceed the Isotig Threshold, further traversal of isotig paths for the group stops.