Size of invariant set

2 views
Skip to first unread message

Lux

unread,
Mar 16, 2011, 4:23:47 PM3/16/11
to Computational Genomics 2011Spring
Here are some backgrounds:
1, every segment has 1000 genes, except the last one, which has 536;
2, the PRD threshold should be non-decreasing
3, each invariant set for each segment should have at least 50 genes

I don't know if anyone has paid attention to the size of invariant
set; for the first ~60 segments, it looks fine; for the last few
segments, which contain large intensity data, the size of invariant
set grows rapidly. In my result the last set has 515 genes, i.e.most
of the genes in the last segment are considered as invariant, but they
have huge difference in array X and Y.

If I set the segment length at 1000, the last regression line has
negative slope; if I set the length at 1024, all lines are non-
decreasing.

Lu

Ziv

unread,
Mar 16, 2011, 4:39:50 PM3/16/11
to Computational Genomics 2011Spring
Hi, its fine to use slightly larger segments as long as you
specifically mention that you are doing this in your solutions.
Another possibility is to fix a value (use a flat line with y=b) if
the last segment (or any segment) is decreasing.
Ziv
Reply all
Reply to author
Forward
0 new messages