[Genome] question about hg18 self-alignment

[Chiang, Charleston]

To whom this may concern,

I was trying to work with the axt files of human hg18 self-alignments (http://hgdownload.cse.ucsc.edu/goldenPath/hg18/vsSelf/axtNet/), and noted that in the few cases I looked, self-aligning chromosomal segments are not reported redundantly.  For example, a segment of chr21 is aligned to a segment of chr15 under the chr21 axt file, yet the same fragment is not reported in the alignment of chr15 to chr21 or any other chromosomal fragments in chr15's axt file.  Similarly, intrachromosomal alignments are also only reported once (i.e. seg A on chr21 aligns to seg B on chr21, but there's no entry of seg B to seg A).  So I would like to make sure that is all alignments reported only once, and in the first scenario I mentioned (fragment of chr21 aligned to fragment of chr15), is the fragment in chr15 not considered for additional alignments once the algorithm has determined that chr21 has aligned to it?  Then the consequence of this method would be that any given segments of the human genome would at most have one matching alignment?

Finally, one more bit of question.  In the axt files, two sets of coordinates are listed (say, chr21 coordinates and chr15 coordinates).  Then if all of first sets of coordinates from all chromosomes together span a certain distance (say, 500 Mb), the actual regions of the genome that can self align is more like 1 Gb?

Thank you for your attention,
Charleston Chiang