Kidney software builds
-
Version 1.0.0
Changes
With the latest release there are now a number of choices available for deployment. The ruby application (called the
kidney_server
) that was distributed in each of the previous releases and which is currently used by NHSBT has had significant updates. Of particular importance is that finding an optimal set of exchanges that contains cycles and chains of sizes 4 and 5 is now feasible and can be achieved quickly with dataset with fewer than 500 patients. These performance improvements can also be seen with lower cycle sizes. However, it should be noted that this application is no longer compatible with Windows and that an additional flag may need to be passed as a parameter if 4-cycles and above are to be found. The details of the new parameters and output format can be found in the REST API guide and the Output Formats respectively. There are no changes required if the length of the largest cycles is 3 and chains of size at most 2 (long chains) are used.In addition to the
kidney_server
application there is also a new web application called thekidney_service
that does not require ruby. This application is of use if the CPLEX solver is required - this is only likely to be the case if the size of the instances increase considerably, for example greater than 500 patients. The details of how to use this application and the installation process are given here. This application does not support Solaris therefore should this application be required then it will be necessary to switch to Linux or Windows. The output data returned by this application is identical to that of thekidney_server
.The final application being made available is a command line application called
kidney_nhs
. There is support for Solaris, Windows and Linux and also for the CPLEX solver with the latter two OSs. This application uses the new column generation techniques that are required should the dataset sizes increase considerably, for example greater than 800 patients. The output data may be of a different format to the web API , details of these changes can be found here. The use of column generation is controlled automatically with the application and can also be switched on and off using config settings.We note that if all that is required is to pick up the latest version of the application with deployment on Solaris then the application can be downloaded here. The deployment guide maybe be useful in this case.
Downloads
kidney_server
kidney_service
kidney_nhs
-
Version 0.0.10
Changes
- Fixed issue with incorrect solution output when
maxcard
option used. Please note: using themaxcard
option is not recommended as it is likely to be deprecated in the future.
Downloads
- Fixed issue with incorrect solution output when
-
Version 0.0.9
Changes
- Updated to new optimality criteria for altruistic donors - i.e. we no longer maximise the number of altruistic donors, see optimality criteria for more details
- Edges from recipients to altruistic donors will have total weight of 0, that is, the age points score
and the tie-breaker score will now both be 0 on an edge of this type
e.g. for such an edge the xml will now be
<tb>0.00000</tb><dif>0</dif>
- Group ids now start from 1 rather than 0
- Version of the server running is now displayed on the bottom right of each page of the web UI
- About link at the top of the page now points to these release notes
Downloads
-
Version 0.0.8
Changes
- Update to force the altruistic donor in the altruistic chain to appear first in the list
Downloads
-
Version 0.0.7
Changes
- Replaced donor-to-donor age difference with donor-to-donor age difference points in the output data
- Added the score passed into the algorithms back out with the output data
Downloads
-
Version 0.0.6
Changes
- Added tie breaker calculation and donor-to-donor age difference to output data
- Added documentation for deploying to Solaris 10 -- deployment guide
Downloads
-
Version 0.0.5
Changes
- Added config file to allow setting log file location, log level and maximum execution time
- Added config file documentation to API Docs
- Server should now be able to start up using
ruby scripts/start -d
option - Significant performance enhancements for larger sized instances
Downloads
-
Version 0.0.4
Changes
- The <b/> elements in the XML output now contains the group id of the embedded pairwise.
- The documentation has been updated to include details of (1) and also contains other minor changes.
- The maxcard optimality criteria has now been changed and now only tries to maximize the number of transplants and subject to this maximize the score.
Downloads