In my efforts to try to debug our installation of the UMLS::Similarity, I ran across this old thread. What causes the following error to appear during the tests
t/create-icfrequency.t .... 1/5 Could not open outputfile t/output/create-icfrequency.output is that in Line 350 of create-icfrequency.pl, it calls count.pl from Text::NSP. If count.pl is not in your path, it fails the create-icfrequency.t test. In our case, count.pl was in /usr/local/bin (installed by MCPAN), but not in root's path. -Albert ---In umls-similarity@yahoogroups.com, <finale.doshi@...> wrote : Thanks for all your help so far. Things were still stalling, so I tried doing a fresh install starting from re-creating the UMLS databases in mysql and installing UMLS-Interface and UMLS-Similarity again. Unfortunately, I think I must have not cleanly uninstalled/something was off because now when I make test on UMLS-Similarity, I get the error below. Suggestions? Make test on UMLS-Interface takes about 20 minutes but passes. Thanks, Finale PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'blib/lib', 'blib/arch')" t/*.t t/create-icfrequency.t .... 1/5 Could not open outputfile t/output/create-icfrequency.output # Looks like you planned 5 tests but ran 1. # Looks like your test exited with 2 just after 1. t/create-icfrequency.t .... Dubious, test returned 2 (wstat 512, 0x200) Failed 4/5 subtests t/create-icpropagation.t .. ok t/error.t ................. ok t/ic.t .................... ok t/lesk.t .................. ok t/path.t .................. ok t/vector-input.t .......... ok t/vector.t ................ ok Test Summary Report ------------------- t/create-icfrequency.t (Wstat: 512 Tests: 1 Failed: 0) Non-zero exit status: 2 Parse errors: Bad plan. You planned 5 tests but ran 1. Files=8, Tests=65, 11608 wallclock secs ( 0.04 usr 0.01 sys + 145.61 cusr 15.40 csys = 161.06 CPU) Result: FAIL Failed 1/8 test programs. 0/65 subtests failed. --- In umls-similarity@yahoogroups.com mailto:umls-similarity@yahoogroups.com, Bridget McInnes <btmcinnes@...> wrote: > > Hi Finale, > > I am glad that solved it. For when you are running umls-similarity.pl (or > any of the programs for the most part) without a configuration file, the > default uses MSH with the PAR, CHD relations. If you run: > > removeConfigData.pl config.txt > > only set the configuration to: > > SAB :: include MSH > REL :: include PAR, CHD > > That should clear it out. > > If you ran with any other configuration settings you might want to > removeConfigData those as well. > > Let us know if you run into any other problems. > > Thanks! > > Bridget > > > > On Wed, Apr 3, 2013 at 7:50 PM, flyingpotatoe42 <finale.doshi@...>wrote: > > > ** > > > > > > I updated my.cnf and ran > > > > removeConfigData.pl config.txt > > > > where config.txt is the file with the SAB::include etc. Now the > > UMLS-Interface make test succeeds in a few minutes, so perhaps there was > > something slowing it down there. > > > > However, now when I even just run something like > > > > perl umls-similarity.pl C0242372 C0004238 > > > > It stalls after "No user defined settings." If I add the --realtime > > option, it stalls after displaying the "PathFinder User Options: realtime > > option set". If I add a config file, it stalls after displaying the > > "CuiFinder User Options: config option set." > > > > Does this mean there's still some stuff somewhere that needs to be removed? > > > > Thanks for all your help, > > Finale > > > > --- In umls-similarity@yahoogroups.com > > mailto:umls-similarity@yahoogroups.com, Bridget McInnes <btmcinnes@> > > wrote: > > > > > > Hi Finale, > > > > > > That is too slow. > > > > > > Did you modify your my.cnf file with the directions from the INSTALL > > file? > > > The UMLS folks have increased the recommended buffer sizes and options > > with > > > one their latest releases. I am behind in updating the INSTALL to reflect > > > this. Here is the link with their recommendations: > > > > > > > > http://www.nlm.nih.gov/research/umls/implementation_resources/scripts/README_RRF_MySQL_Output_Stream.html > > > > http://www.nlm.nih.gov/research/umls/implementation_resources/scripts/README_RRF_MySQL_Output_Stream.html > > > > > > If those haven't been updated, I would suggestion halting the program, > > > modifying the my.cnf file and then re-start. Prior to re-starting though, > > > you need to remove the index file in the umlsinterfaceindex that has been > > > generated: > > > > > > removeConfigData.pl <configuration file> > > > > > > If you have modified the my.cnf file with the new sizes ... I need to > > > think. Let me know if you have and do you remember which tests you > > stalled > > > out on with UMLS-Interface? > > > > > > Thanks! > > > > > > Bridget > > > > > > > > > > > > On Wed, Apr 3, 2013 at 2:00 PM, flyingpotatoe42 <finale.doshi@>wrote: > > > > > > > ** > > > > > > > > > > > > Thanks for your prompt reply and detailed instructions! > > > > > > > > I tried checking the number of entries in table associated with > > > > > > > > MMSYS_2012AB_20121025_SNOMEDCT_CHD_PAR_cache > > > > > > > > and there are only 5 entries, after running for a day! Do you have any > > > > suggestions of what I might need to change or optimize? I am running > > with > > > > UMLS-2012AB, SAB :: include SNOMEDCT, and REL :: include PAR, CHD. > > > > > > > > Not sure if this is related: when I ran "make test" after installing > > > > UMLS-Interface, the tests would always succeed until they stalled at > > some > > > > point (different points). I ignored this because "make test" on > > > > UMLS-Similarity succeeded quickly, but I didn't know if that helps > > give an > > > > indication on why this script seems to be stalling. > > > > > > > > Thanks! > > > > Finale > > > > > > > > --- In umls-similarity@yahoogroups.com > > > > mailto:umls-similarity@yahoogroups.com, Bridget McInnes <btmcinnes@> > > > > wrote: > > > > > > > > > > Hello Finale, > > > > > > > > > > The initial index will take a while -- it is a little hard to > > determine > > > > how > > > > > long because it depends on your system. > > > > > > > > > > One thing I normally do when initially creating an index is to run it > > > > first > > > > > with a small example, then on the larger set. I will add an index > > > > finished > > > > > comment to the next release -- this will help determine at what > > stage of > > > > > the process you are at. I will also look at adding some more > > information > > > > to > > > > > help gauge the status of how far along you are. > > > > > > > > > > You can check to see if/how fast the table is filling up in the > > > > > umlsinterfaceindex. I am not certain how familiar you are with mysql > > so > > > > > below I describe what I normally do: > > > > > > > > > > Sign into to mysql: > > > > > > > > > > mysql -u <username> -p > > > > > > > > > > this will prompt you for your password. Next type the following: > > > > > > > > > > use umlsinterfaceindex; > > > > > select * from tableindex; > > > > > > > > > > find where it says something along the lines of: > > > > > MMSYS_2012AA_20120322_SNOMEDCT_CHD_PAR_cache > > > > > > > > > > this is dependent on your UMLS version, and the SAB/REL options in > > the > > > > > configuration file > > > > > > > > > > Copy that alpha-numeric sequence next to it. > > > > > > > > > > then type: > > > > > > > > > > select count(*) from <alpha-numeric sequence>; > > > > > > > > > > This will provide you the number of lines in the umlsinterfaceindex > > for > > > > > that configuration. The count(*) when using > > > > > UMLS-2012AA > > > > > SAB :: include SNOMEDCT > > > > > REL :: include PAR, CHD > > > > > > > > > > is 13,246,070 > > > > > > > > > > That is how I tend to monitor the progress. I hope this helps some. > > > > > > > > > > Thanks, > > > > > > > > > > Bridget > > > > > > > > > > > > > > > On Wed, Apr 3, 2013 at 7:52 AM, flyingpotatoe42 <finale.doshi@ > > >wrote: > > > > > > > > > > > ** > > > > > > > > > > > > > > > > > > Thanks for the pointers! I installed UMLS Similarity locally and > > set my > > > > > > batch of ~6000 concepts running with the infile option. It's been > > > > running > > > > > > for about a day now (measure wup, sab SNOMEDCT, yes to building the > > > > index). > > > > > > Is this to be expected? > > > > > > > > > > > > I don't mind waiting a few days; mostly I just wanted to get a > > sense of > > > > > > how long I should expect it to take before I try any tuning that > > you > > > > > > alluded to in your email. > > > > > > > > > > > > Thanks, > > > > > > Finale > > > > > > > > > > > > --- In umls-similarity@yahoogroups.com > > > > > > mailto:umls-similarity@yahoogroups.com, Ted Pedersen <duluthted@> > > > > > > wrote: > > > > > > > > > > > > > > There are probably two things you can do - first, if you need to > > run > > > > a > > > > > > > lot of queries, you'll want to install UMLS::Similarity locally. > > The > > > > > > > web interface is only set up to one a single query at a time, so > > that > > > > > > > will always be a bit slow. > > > > > > > > > > > > > > Second, once you have UMLS::Similarity you can run use the > > --infile > > > > > > > option with umls-similarity.pl. This lets you load a file of CUI > > > > > > > pairs. This is faster than running each query individually since > > the > > > > > > > setup overhead is only an issue when umls-similarity.pl starts. > > > > Also, > > > > > > > with UMLS::Similarity installed locally you can also do some > > tuning > > > > of > > > > > > > the UMLS which can also help performance. > > > > > > > > > > > > > > I hope this helps. Please feel free to follow up with any > > additional > > > > > > questions! > > > > > > > > > > > > > > Good luck, > > > > > > > Ted > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >
