Order contig & position correctly

[ihodes]

We're ALMOST there:

We sort by contig length, and then lexicographically. We need to do a little better…

[danvk]

We'd get the desired behavior if we sorted a zero-padded version of the
contig name. You'd have to be careful to add the right # of zeros to just
the numbers, though (e.g. "2"→"02" but "20"→"20" and "X"→"X"). Not sure how
easy this is in SQLAlchemy/Postgres.

On Wed Dec 17 2014 at 12:37:03 PM Isaac Hodes notifications@github.com
wrote:

We're ALMOST there:

https://camo.githubusercontent.com/f415d8c8c07bc30fbfb6d2d7d9e7b6bdb299103c/68747470733a2f2f73332e616d617a6f6e6177732e636f6d2f662e636c2e6c792f6974656d732f32353343304f3270337730573241324f326530362f53637265656e25323053686f74253230323031342d31322d3137253230617425323031322e33362e3033253230504d2e706e67

We sort by contig length, and then lexicographically. We need to do a
little better…

—
Reply to this email directly or view it on GitHub
#407.

[arahuja]

Not sure if this the same issue or a new one, but looks like positions are not sorted correctly now either:

From this run

[ihodes]

FWIW, while I work on the issue + translate this into SQLAlchemy, the SQL to sort contig/position nicely looks like:

order by COALESCE(SUBSTRING(contig FROM '^\d+')::INTEGER, 1000), length(contig), contig, position;

[ihodes]

This regressed somehow and passed tests: http://cycledash.demeter.hpc.mssm.edu/runs/153/examine

Need to fix and add better regression testing.

[ihodes]

This appears to only be a problem when contigs are prefixed with chr.