Some months ago, I brought this up incorrectly. I've spoken to the
scientists here, and I'd like to propose a set of semantics for
BetweenLocations. This is driven by the two genbank/embl cases 5^6 and
5^10. LocationTools will require the boolean operations areEqual, contains,
and overlaps. It also requires operations that return a location, union and
intersection.
The general proposal is to be generous with returning values where the
correct behavior is ill-defined. For example, we recommend 5^10 intersects
7^12 results in 7^10. It's easiest to define the semantics for each case.
Assumptions
* Union, areEqual, Overlaps, and Intersection are symmetric operators
* Contains is a directional operator
* In general, 7^12 is analogous to 7.12 in behavior
* A location 1..10 represents nucleotides 1 through 10 inclusive
* A location 5^6 represents the space between nucleotide 5 and nucleotide 6
* A location 5^10 represents a space between two nucleotides. The
nucleotides are located between 5 and 10 inclusive
* Requesting the features on a subsequence must return the features with
between locations as well
(tab-delimited table)
Location Intersection Overlaps
1..10 intersects 5^6 result 5^6 TRUE
1..10 intersects 5^7 result 5^7 TRUE
1..10 intersects 10^11 result EMPTY FALSE
1..10 intersects 9^11 result 9^10 TRUE
5^6 intersects 5^6 result 5^6 TRUE
5^10 intersects 5^6 result 5^6 TRUE
5^10 intersects 7^12 result 7^10 TRUE
areEqual
5..6 equals 5^6 result FALSE
5^6 equals 5^6 result TRUE
5^6 equals 5^10 result FALSE
Contains
1..10 contains 5^6 result TRUE
5^6 contains 1..10 result FALSE
1..10 contains 5^7 result TRUE
1..10 contains 10^11 result FALSE
1..10 contains 9^11 result FALSE
5^6 contains 5^6 result TRUE
5^10 contains 5^6 result TRUE
5^6 contains 5^10 result FALSE
5^10 contains 7^12 result FALSE
7^12 contains 5^10 result FALSE
Overlaps
1..10 overlaps 5^6 result TRUE
1..10 overlaps 5^7 result TRUE
1..10 overlaps 10^11 result FALSE
1..10 overlaps 9^11 result TRUE
9^11 overlaps 1..10 result TRUE
5^6 overlaps 5^6 result TRUE
5^6 overlaps 5^10 result TRUE
5^10 overlaps 7^12 result TRUE
Union
1..10 union 5^6 result compoundLocation(1..10,5^6)
1..10 union 5^7 result compoundLocation(1..10,5^7)
1..10 union 10^11 result compoundLocation(1..10,10^11)
1..10 union 9^11 result compoundLocation(1..10,9^11)
5^6 union 5^6 result 5^6
5^10 union 5^6 result 5^10
5^10 union 7^12 result 5^12
If there aren't any objections, I'll code this up and commit it.
Greg
_______________________________________________
Biojava-l mailing list - [EMAIL PROTECTED]
http://biojava.org/mailman/listinfo/biojava-l
