Take the union of GN / VL parsing and geocoding set from BioSamples

[ababaian]

We have two datasets that have "passed" over the BioSample XML files, there are some discrepancies between them, we should take the union of the parsed coordinates.

Set 1: VL geocoded_all is on the SQL server.

Set 2: GN df_final_lat_lon.csv which is the product of some manual labour. Download from S3: s3://serratus-public/tmp/df_final_lat_lon.csv

[victorlin]

VL visual:

GN visual:

Noticable differences in the marinal points.

[ababaian]

Intersection between the GN and VL sets:

> length(VL)
[1] 2534331
> length(which(VL %in% GN))
[1] 1748525
> length(VL) - length(which(VL %in% GN))
[1] 785806

> length(GN)
[1] 1478818
> length(which(GN %in% VL))
[1] 1465216
> length(GN) - length(which(GN %in% VL))
[1] 13602

When I plot your geocoded data, it gives me the same plot as above, it looks like the difference is things are missing from the plotting in the geocoded_all dataset, but the data is there?