Refactoring FASTA work to break contig sizes. #160
Conversation
| union { null, long } sequenceLength = null; | ||
| union { null, string } url = null; | ||
| union { null, string } url = null; | ||
| array<Base> fragmentSequence = []; // sequence of bases in this fragment |
There was a problem hiding this comment.
Was there some discussion of moving away from array to string to represent a sequence?
There was a problem hiding this comment.
@arahuja, I believe @nealsid suggested it a while back. I can run an experiment in the next few days to check this; I believe that Neal was concerned that we would get better compression if we used strings instead of an array of enums. This should be a straightforward change to make and an easy one to evaluate.
I think the advantage you do get from defining your alphabet via an enum is that you are strictly checking for possible errors; whether this is valuable is another question.
There was a problem hiding this comment.
@fnothaft Should we use a string here?
|
One or more automated tests failed |
|
Jenkins, test this please. |
|
One or more automated tests failed |
|
Jenkins, test this please. |
|
All automated tests passed. |
|
Jenkins, test this please. |
|
One or more automated tests failed |
|
Jenkins, retest this please. |
|
All automated tests passed. |
|
Jenkins, test this please. |
|
One or more automated tests failed |
|
Looks like an issue with an implicit conversion... Fixing now. |
|
All automated tests passed. |
Refactoring FASTA work to break contig sizes.
|
Thanks, Frank! |
This change fixes #109. Specifically, on FASTA import, assemblies that were too long (>800kbp) would cause things to go a bit wonky. To resolve this, we break a single contig up into contig fragments. Additionally, this PR adds a few utility functions for manipulating contigs.
As a note, I've held off on editing the CHANGES.txt/md until there is consensus on which one we are sticking with.