Hi, yes the problem is that at the end of the code the index is not builded completely and the removeConfigData.pl is not working (see the error in previous message). I had this problem while I was using an UMLS subset with only two semantic types obtained using Metamorphosis. After noticing the first error I tried to build an index over the entire UMLS and this time the program seems to never end. I was also keeping track of the queries and in my general log table there where more than 5 million queries on umls and umlsindex databases. After your message I stopped the execution to add the realtime option as you suggested. As before I’m logging the queries. Actually after more than one hour I can count about 1800000 selects on cuis and the program is still running. My configuration file is simply SAB :: include UMLS_ALL REL :: include PAR, CHD
while the other parameters (database, password, socket…, and now realtime) are specified in the code as you can see in the previous message. Thanks for your help, Eugenia Eugenia Galeota, PhD Center for Genomic Science of IIT@SEMM Computational Epigenomics Email: [email protected] Tel: +39 02 9437 5046 Via Adamello 16, 20139 Milan, Italy On 08 Oct 2014, at 14:33, Bridget McInnes [email protected] [umls-similarity] <[email protected]> wrote: > > Hi Eugenia, > > I apologize that things are not running smoothly. I would like to verify a > few things with you: > > 1. the problem is that the index will not build; and removing the index using > the removeConfigData.pl is not working? What configuration file where you > using for this? > > > 2. you are running the index over the entire UMLS? If so, we recommend that > you use the --realtime option if you would like to take the similarity scores > between concepts using the path information from the entire UMLS. The UMLS is > so large, that there is typically not enough space to store all the > path-to-root information for all the CUIs. The --realtime option allows the > path information to be collected on the fly. It takes a little longer though; > so we created the index for smaller sources/relation configurations and the > --realtime option for the larger configurations. I hope that makes sense. > Please let me know if it does not. > > Thank you! > > Best regards, > > Bridget > > > On Wed, Oct 8, 2014 at 7:09 AM, Eugenia Galeota [email protected] > [umls-similarity] <[email protected]> wrote: > > > Dear developers, > I’ve been trying in many ways to let UMLS::Similarity work, but it seems > that there are some problems I’m not able to solve. Initially I had issues > with my.cnf and mysql.socket. I had different versions of the same files, > then the mysql.socket disappeared from the /tmp folder. Actually they seems > to be solved, since I completely removed any previous mysql server > installation and reinstalled the latest version using brew on my mac. I was > using UMLS similarity on a small subset with only two semantic types created > using Metamorphosys. I tested some of the Interface methods and they seems to > be working, anyway when trying to create an index to use with umps similarity > I always get the following error > > ERROR: UMLS::Interface::PathFinder->_checkIndex > Index Error (Error Code 9). > Index did not complete. Remove using the removeConfigData.pl program and > re-run. > > and when I try to use the removeConfigData.pl to rebuild it from scratch the > program never ends (it is still running since last friday). > I thought that another possibility is that in some way the Pathfinder module > cannot build the paths to the root of the UMLS because tables in mysql don't > contain the UMLS root concept C0000000 and its relationships. By looking at > the code I still wasn’t able to understand how the path from concepts without > a parent to the UMLS root is created. I found a call at a certain point to > some function getCuiChildren where the passed cui string seems to be an empty > string empty. I also modified the my.cnf file as you specify in the > documentation of the perl module. > Finally I’m trying to do the same operations on the entire UMLS database. My > code is the following > > use lib '/Library/Perl/5.16/'; > use UMLS::Interface; > use UMLS::Similarity::lin; > %params = (); > $params{"intrinsic"} = "sanchez"; > $params{"username"} = "egaleota"; > $params{"password"} = “password"; > $params{"socket"} = "/tmp/mysql.sock"; > $params{"config"} = "/Users/egaleota/umls/UMLS-Interface-1.41/myconfig.conf"; > #$cuifinder = UMLS::Interface::CuiFinder->new(\%params); > my $umls = UMLS::Interface->new(\%params); > die "Unable to create UMLS::Interface object.\n" if(!$umls); > $paramtri{"intrinsic"}="sanchez"; > my $lin = UMLS::Similarity::lin->new($umls, \%paramtri); > #die "Unable to create measure object.\n" if(!$lin); > my $cui1 = "C0459385"; > my $cui2 = "C0440746"; > $ts1 = $umls->getTermList($cui1); > my $term1 = pop @{$ts1}; > $ts2 = $umls->getTermList($cui2); > my $term2 = pop @{$ts2}; > my $value = $lin->getRelatedness($cui1, $cui2); > print "The similarity between $cui1 ($term1) and $cui2 ($term2) is $value\n"; > > > > Actually I’m running the code above. The index seems to be in creation phase. > When I check for the current queries in the db I obtain results like the > following which let me believe that the problems are not related to how perl > communicates with mysql > > +----+----------+-----------+--------------------+---------+------+------------+------------------------------------------------------------------------------------------------------+ > | Id | User | Host | db | Command | Time | State > | Info > | > +----+----------+-----------+--------------------+---------+------+------------+------------------------------------------------------------------------------------------------------+ > | 51 | egaleota | localhost | umls | Query | 0 | > statistics | select distinct CUI2 from MRREL where CUI1='C3135343' and > ((REL='CHD') ) and CUI2!='C3135343' and SU | > | 52 | egaleota | localhost | umlsinterfaceindex | Sleep | 0 | > | NULL > | > | 54 | egaleota | localhost | NULL | Query | 0 | init > | show processlist > | > +----+----------+-----------+--------------------+---------+------+------------+------------------ > > > Since all this tasks are really time consuming I was kindly wondering if you > could help me to solve this issues. Thanks, > > Eugenia Galeota, PhD > Center for Genomic Science of IIT@SEMM > Computational Epigenomics > Email: [email protected] > Tel: +39 02 9437 5046 > Via Adamello 16, 20139 Milan, Italy > > > > >
