33 lines
1.2 KiB
Text
33 lines
1.2 KiB
Text
1001 Ways to Make COP Better
|
|
|
|
Simon Dear
|
|
3 December 1991
|
|
|
|
|
|
1. Incorporate quality considerations.
|
|
Certain problems can slip through the detetion net. An example are
|
|
overcalled bases in the consensus. Currently the whole raw data read
|
|
is used, and overcalling will agree with the consensus. Overcalled
|
|
runs of bases are very symptomatic of poor trace quality. If these
|
|
could be detected and excluded from the coverage map, the problem
|
|
would be detected.
|
|
|
|
2. Running the program on contigs and regions of contigs.
|
|
Cop could then be run on partially completed project databases.
|
|
|
|
3. Provide explanations as to why a region is problematic.
|
|
Maybe this isn't worth the effort to implement, but would be useful.
|
|
|
|
4. Take into account strandedness.
|
|
This would help detect regions where the coverage by good quality
|
|
traces is poor.
|
|
|
|
5. Interface to contig editor.
|
|
The contig editor could center on and highlight regions where problems
|
|
exist, speeding up the process.
|
|
|
|
6. Consensus coverage.
|
|
Currently if 10 good readings disagree with a region and just 1
|
|
mediocre reading agrees, no problem is detected.
|
|
|
|
7-1001. Make it detect 995 times as many problems.
|