2007AB

Title: Validation Error when running off DVD

Problem: Users who have a DVD read/write drive will receive this error when attempting to validate the UMLS 2007AB DVD.

Solution: Until the user clicks "OK" the installation process can not go forward. Clicking "OK" will allow the validation to complete by writing the file to another location.

Date posted: July 2, 2007

2007AB

Title: Non-ASCII UTF-8 dash characters in 5 RxNorm, MTHSPL CODES not removed by MetamorphoSys Output Option

Problem:

Who is affected? Users who
 (1) include RxNorm or MTHSPL data in their customized Metathesaurus subsets, and who also
 (2) use the Output Option in MetamorphoSys to exclude non-UTF-8 characters from their subset

MetamorphoSys fails to remove the following non-ASCII UTF-8 dash characters in the CODE field in the following 5 rows in MRSAT (2007AB):
 
C0716197|L6401119|S7339522|A11769224|AUI|0025–1961|AT69824745||NDC|MTHSPL|0025-1961-30|N||
C0716197|L6401119|S7339522|A11769224|AUI|0025–1961|AT69829041||SPL_SET_ID|MTHSPL|0B0BA593-FE54-456C-A3D7-651C 50036E3E|N||
C0716197|L6401119|S7339522|A11769224|AUI|0025–1961|AT69831407||RXCUI|RXNORM|687204|N||
C0716197|L6401119|S7339522|A11769224|AUI|0025–1961|AT69851217||DM_SPL_ID|MTHSPL|2746|N||
C0716197|L6401119|S7339522|A11769224|AUI|0025–1961|AT69853512||RXAUI|RXNORM|2621866|N||

It fails to remove the non-ASCII UTF-8 dash in in the CODE field of one MRCONSO row:

C0716197|ENG|S|L6401119|PF|S7339522|Y|A11769224||||MTHSPL|DP|0025–1961|Metronidazole 750 MILLIGRAM In 1 TABLET ORAL TABLET, FILM COATED, EXTENDED RELEASE [Flagyl ER]|0|N||

The dash character in these rows is not an ASCII dash, but rather the UTF-8 dash: 
0x2013|U+2013|EN DASH|General Punctuation|1
 

Solution:

The Unicode character U+2013 is represented in UTF-8 as a sequence of the following 3 bytes (in octal):
\342, \200 and \223.

Use the following perl script to make the change in the appropriate file:
perl -ne '{s/\342\200\223/-/g;print}' MRCONSO.RRF
or
perl -ne '{s/\342\200\223/-/g;print}' MRSAT.RRF

These errors will be corrected in the 2007AC release.

Date posted: July 2, 2007

2007AA

Title: Bad LOINC Relationship data in early release

Problems:
Relationship data was incorrect in an early edition of the 2007AA release. Users who downloaded the release prior to the official announcement will find incorrect LOINC mapped-to relationships in the file MRREL.

Solution:
The information has been corrected in the official 2007AA release. Users should download these files which were made available on 1/29/07.

Date posted: January 30, 2007

2007AA

Title: Oracle Load Script produces errors related to Japanese language sources

Problems:
Oracle users receive errors when trying to load the file RXW_JPN.RRF.

Solution:
The log file will report a series of errors like the one below:

Record 6171: Rejected - Error on table MRXW_JPN, column WD.
Field in data file exceeds maximum length

The error occurs because Oracle interprets the length of the WD column of the MRXW_JPN table in terms of BYTE semantics, instead of CHAR semantics.  To force SQL*Loader to handle this properly, the MRXW_JPN.ctl SQL*Loader control file can be edited to include this line:

characterset UTF8 length semantics char

Thus, the complete MRXW_JPN.ctl file would be as follows:

options (direct=true)
load data
characterset UTF8 length semantics char
infile 'MRXW_JPN.RRF' "str X'7c0a'"
badfile 'MRXW_JPN.bad'
discardfile 'MRXW_JPN.dsc'
truncate
into table MRXW_JPN
fields terminated by '|'
trailing nullcols
(LAT    char(3),
WD      char(100),
CUI     char(8),
LUI     char(8),
SUI     char(8)
)

With this change, the populate_oracle_db.csh script should run to completion without any errors.

Date posted: January 30, 2007

2007AA

Title: Oracle Load Script produces errors related to character fields

Problems:
Oracle users receive errors when trying to load the file MRSAT.

Solution:
Set "Truncate long fields to 3990 characters" to avoid errors in loading MRSAT rows into Oracle. "Truncate long fields..." appears on the "Output Options" tab, under "Maximum Field Length". This problem will be fixed by changing the default field length in the 2007AB Release.

Date posted: January 15, 2007

2007AA

Title: MRHIER and MRSAB don't load properly in Oracle databases

Problems:
Oracle users receive errors when trying to load the files MRSAB & MRHIER.

Solution:

If you have already created a subset, edit the file "oracle_tables.sql" or "mysql_tables.sql" so that the TTYL field has a length of 200 (instead of 100) and the PAUI field has a VARCHAR (or in Oracle VARCHAR2) data type with length 9 (instead of CHAR(8)). If loading Oracle scripts, also edit the SQL*Loader control files "MRSAB.ctl" and "MRHIER.ctl" to accordingly set the lengths of the TTYL and PAUI fields.

Prior to creating a subset:

Edit the MMSYS/config/mrpluscolsfiles.dat file. Change the lines

TTYL|MRCOLS|varchar(100)|Term type list for a source||||06-DEC-06|
PAUI|MRCOLS|char(8)|Unique identifier for parent atom||||06-DEC-06|

To this:

TTYL|MRCOLS|varchar(200)|Term type list for a source||||06-DEC-06|
PAUI|MRCOLS|varchar(9)|Unique identifier for parent atom||||06-DEC-06|

This problem will be fixed in the 2007AB Release.

Date posted: January 15, 2007

2007AA

Title: Macintosh directory names with spaces

Problems:
On Macintosh OS X, the name of the install directory and its parents can not include spaces; shell scripts will fail to find the current directory name, and cause this error message to appear:

cd: MMSYS: No such file or directory

Solution:
Install the UMLS release files in an install directory with no spaces in its name or the names of its parents.

Date posted: January 15, 2007

2006AC, 2006AD

Title: MetamorphoSys Content View Filter Tab 

Problems:
1) Enabling the Content View Filter creates a blank tab. The Natural Language Processing Content View Filter is not displayed and cannot be selected.
(2) Users are unable to navigate away from the blank tab.  (Temporary and immediate solution:  Disable the Content View Filter.)

Solution:
The error exists in the files MMSYS/config/mmsys.a.prop and MMSYS/config/mmsys.b.prop.

Add the lines below to the both of both files:  

gov.nih.nlm.mms.filters.ContentViewFilter.name_2=MetaMap NLP View
gov.nih.nlm.mms.filters.ContentViewFilter.description_2=It represents the strict model of data used by MetaMap.
gov.nih.nlm.mms.filters.ContentViewFilter.code_2=256

Adding these lines will correct the error and allow users to enable a Content View Filter that correctly displays the Natural Language Processing filter. 

Date posted: November 3, 2006

2006AC

Title: Advanced Suppressibility Options -->Remove Suppressible Term Types Option does not remove Term Types

Problem:
Users who create an ORF subset and select one or all of the following options: Remove Source Term Type Suppressible Data, Remove Editor Assigned Suppressible Data, or Remove Obsolete Data, will find the line "Suppressible terms were included in the subset." at the bottom of their MMSYS.log file.

Solution: The error exists in the production of the log file, not in the data files. Examination of the data files will show that suppressible terms were not included in the subset, in relation to the options the user checked. (For example, if Remove Obsolete Data was checked only obsolete terms will be removed.) The error in the log file will be corrected for the 2006AD release.

Date posted: July 14, 2006

2006AC

Title: In the RRF Browser, Find command does not highlight terms

Problem:
On Mac, Solaris, and Linux platforms: when using the find command, found terms are not highlighted until the user moves (scrolls up or down) within the window.

Solution: The error will be corrected for the 2006AD release.

Date posted: July 14, 2006

2006AB

Title: Java Issues After Installing J2SE 5.0 Release 4 Update

Problem: The software update for Java 2 Platform Standard Edition (J2SE) 5.0 Release 4 for Mac OS X v10.4.5 will change the default version of Java virtual machine from Java 1.4.2 to Java 5.0. This may cause problems with MetamorphoSys.

Solution: Reset the default Java virtual machine. Instructions are available from Apple at http://docs.info.apple.com/article.html?artnum=303526. The 2006AC release of the UMLS will call the correct JRE, regardless of the default settings.

Date posted: 5/08/06

2006AA

Title: Incorrect MeSH^® Relationships

Problem: In the UMLS 2006AA release available for download from 2/13/06 to 2/24/06, erroneous relationships were present in the MRREL file and duplicated in the files MRHIER, MRFILES, and MRCOLS.

Solution: A corrected release was made available for download from the UMLSKS on 3/2/06.

Date posted: 2/24/2006

2006AA

Title: Problem with MRHIER.RRF References

Problem: The removal of MRCXT.RRF from the default data set for 2006AA had the unintended consequence of affecting hierarchy information in MRHIER.RRF, such that the information cannot be properly resolved by MetamorphoSys in all cases.

In particular, in cases where a source with hierarchies is selectively subsetted, a situation may result where the PTR field of MRHIER.RRF contains references to AUI values that do not exist in MRCONSO.RRF.

Scope: This will likely not be an issue for the majority of users. In the event that a user's configuration causes MetamorphoSys to either keep or remove sources in their entirety, this problematic data condition does not arise.

There are four known ways in which this condition might occur. Each involves removing parts of sources that contain hierarchies without removing the entire source and operating with an input set that does not contain MRCXT data:

1. Using the "remove UTF8" option in the output handler configuration panel in conjunction with a source list that included sources with UTF-8 characters and a hierarchy
2. Selecting "remove suppressible data" from the suppressible termgroups advanced options and choosing suppressible termgroups from among the atoms of a source that contains hierarchies
3. Using the SampleMetaFilter (very rare)
4. Using a user-created filter that selectively removes atoms from sources with hierarchies (very rare)

The first two situations will by far be the most common.

When using the "remove UTF8" condition in conjunction with a Source List configuration that includes sources with UTF-8 characters, the resulting MRHIER.RRF will contain this data condition. The set of sources containing UTF-8 characters (and contexts) includes:

When selecting term types to be suppressible that participate in contexts and using the advanced options "remove suppressible data" option, the resulting MRHIER.RRF will contain this data condition.

Following is a sample problematic MRHIER.RRF entry that occurs when the user configures MetamorphoSys to "remove UTF-8" and to keep all sources.

 
C0000039|A7439429|1|A7455821|MSHFRE||A7474418.A7456206.A7456207.A7461322.A7452688.A7440176.A7452689.A7455821|||

In this example, AUIs A7452688, A7452689, and A7455821 are not found in MRCONSO.RRF.

Here is another example:

 C0000139|A7480712|1|A7504811|MSHGER||A7533768.A7484783.A7525302.A7504817.A7504811|||

In this case, AUI A7525302 is not found in MRCONSO.RRF.

Solution: There is a work-around for this problem, albeit a time-consuming one. As indicated above, this problem results when the input set does not contain MRCXT.RRF data. The presence of MRCXT.RRF on the input side would guarantee output data free of this data condition.

The user should take the following steps:

1. Generate an output set using the desired configuration, making sure not to select options matching any of the four conditions described above. The output subset should then contain all of the hierarchy data for the set of sources selected for inclusion.
2. Use the MRCXT builder to create a MRCXT.RRF file matching the MRHIER.RRF file.
3. Use the first output set (including MRCXT.RRF) as the input and create a second output, this time using the options that created the initial problem. This second output set should be what the user is looking for.

So, consider the case where the user wanted the "keep everything" subset with the "remove UTF-8" option selected. In order to achieve this via the work-around, the user could take the following actions:

1. Configure the Source List to include all sources.
2. Ensure that "remove UTF-8" was NOT checked in the output handler configuration.
3. Generate the output set.
4. Run the MRCXT Builder to generate a full MRCXT.RRF file.
5. Re-configure MetamorphoSys to point to the set of data generated in #1-4. Again include all sources but this time make sure that "remove UTF-8" IS checked.
6. Generate a second output set. This one will contain correctly resolved MRHIER.RRF data.
7. Remove the first set of output files (unless additional subsets are to be made from it).

A more technically sophisticated user could also use a perl script like the one below to manually remove the unwanted MRHIER.RRF entries from their subset.

	 #!/usr/local/bin/perl

open (MRCONSO,"MRCONSO.RRF") || die "Could not open MRCONSO.RRF : $! $?\n"; while (‹MRCONSO›) {
  chop;
  split /\|/;
  $auis{$_[7]} = 1;
}
close (MRCONSO);
open (MRHIER,"MRHIER.RRF") || die "Could not open MRHIER.RRF : $! $?\n"; while (‹MRHIER›) {
  chop;
  split /\|/;
  $ptr = $_[6];
  @ptr_auis = split /./, $ptr;
  $missing = 0;
  foreach $aui (@ptr_auis) {
     unless ($auis{$aui}) {
       $missing = 1;
       last;
     }
  }
  print "$_\n" unless $missing;
}
close (MRHIER);
	 

The output of this script would be a corrected MRHIER.RRF file.

Date posted: 2/27/2006

2006AA

Title: LSN Attribute Incorrectly Defined

Problem: The LSN attribute is incorrectly defined as "LOINC short name".

Solution: The correct definition for this attribute is "LOINC Spanish name". The offical documentation is correct and can be accessed at http://www.nlm.nih.gov/research/umls/documentation.h tml. This issue will be resolved in the 2006AB release.

Date posted: 2/13/2006

2005AC

Title: ATN Field Too Short in the ORF File MRCOLS

Problem: When generating an ORF subset with the database load script option, the ATN field is set to 20 chars instead of 50.  Any sources that have ATN values greater than 20 that are loaded into a database using one of our scripts produces an error (data is longer than the field length).

Solution: After generating subset and the load script, edit the load script to make ATN field 50 chars. This issue will be resolved in the 2006AA release.

Date posted: 11/30/2005

2005AC

Title: TS Fields in MRCON Not Properly Computed for Editor Suppressed Atoms

Problem: When generating an ORF subset from ORF input, the TS fields in MRCON are not properly computed for editor suppressed atoms.

Solution: This issue will be resolved in the 2006AA release.

Date posted: 11/30/2005

2005AB

Title: MeSH Attribute Information Incorrectly Represented

Problem: The data problem reported on 7/19/2005 has been fixed. In the original 2005AB release (release date 20050615), some data in the following MeSH attributes were incorrectly represented in the Metathesaurus files MRSAT and MRREL: 

Attributes from other sources, and MeSH data in other Metathesaurus files, including MRCONSO, MRDEF, MRSTY, and MRHIER, were unaffected.

Solution: Licensees who use MeSH attribute data in their research or applications should use the corrected 2005AB UMLS release. The corrected 2005AB UMLS release carries the release date of 20050718. Release date is included in the release.dat and mmsys.log files.

Date posted: 7/26/05

2005AB

Title: MetamorphoSys Errors Associated with Mac OS Tiger

Problem: We have found issues in the most recent Tiger update (10.4.2) that may cause new problems for users, including application crashes. We are investigating and expect Apple to remedy these issues. One specific problem: Mac OS Tiger users receive an error message stating "Resources temporarily unavailable" when running MetamorphoSys.

Solution: This error has been corrected in the 2005AC release of the UMLS. If you must use an older version, close all unnecessary applications. Macintosh Tiger users (OS X 10.4) are reminded to quit all other applications (including dashboard widgets) to avoid the Mac "resource temporarily unavailable" error.

Date posted: 7/27/05

2005AB

Title: Differing MD5 Files for AMBIGLUI.RRF, AMBIGSUI.RRF, MRAUI.RRF, MRCUI.RRF, MRSMAP.RRF, and MRSMAP.RRF

Problem: MD5 files created using Windows are not the same as MD5 files created on other operating systems.

Solution: There is no solution at this time. This problem will be corrected in the 2005AC release.

Date posted: 6/15/05

2005AB

Title: Disk Full Errors Are Not Detected and Reported in MetamorphoSys

Problem: In some circumstances disk full errors are not properly detected. MetamorphoSys may appear to complete correctly but files are corrupted. Files may include truncated lines or null characters inserted into the data.

Solution: Users must be certain that they have adequate disk space before using MetamorphoSys.

Date posted: 6/15/05

2005AA

Title: Invalid Data in NDDF Update; Also Affects RxNorm Subsets

Problem: There are serious errors in the NDDF update in 2005AA that affect not only NDDF in the Metathesaurus, but also RxNorm. Drug data that was added to the Metathesaurus from the November NDDF update are correct, but over 100,000 NDDF drug names already in the Metathesaurus were incorrectly marked as obsolete. NDDF data, as a separate source, and within the RxNorm subset, should not be used, or should be used with extreme caution.

Solution: A new RxNorm Cumulative Release File, including corrected NDDF data, will be available in late February or early March.
NDDF will be corrected in the UMLS 2005AB release, scheduled to be available in April 2005.
See also: RxNorm Cumulative Update.

Date posted: 2/8/2005

2005AA

Title: RxNorm Cumulative Update: Non-Matching Codes for RxNorm Atoms

Problem: The RxNorm Cumulative Update of 11/17/2004 contains non-matching codes for RxNorm atoms. This update is based on the 2004AC release of the UMLS, but contains all new codes for RxNorm atoms. These codes do not correspond to previously assigned codes and the update cannot be applied to existing RxNorm data.

Solution: RxNorm data will be resynchronized with the 2005AA release, updated with SNOMED CT drug data, and the full and corrected Cumulative RxNorm will be released in late February or early March.

Date posted: 2/8/2005

2005AA

Title: Missing Codes in Spanish and Portuguese Translations of MeSH 2005 (MSHSPA and MSHPOR)

Problem: For 2005AA, the Spanish and Portuguese translations of MeSH 2005 (MSHSPA and MSHPOR) do not include the MeSH Descriptor Number in the CODE field in MRCONSO.RRF (MRCON in ORF).

Solution: Problem will be corrected in 2005AB.

Date posted: 1/18/2005

2004AC
There were no known issues for 2004AC.

2004AB

Problem: The Original Release Format (ORF) MySql and Oracle load scripts for indexes are missing column names.

1. Manually insert column names into script for indexes
   or
2. Download and unzip the mmsys.zip file (64364251 bytes or 61 MB), then re-run MetamorphoSys to recreate your data
   
   

2004AB

Condition: User selects RxNorm subset from the initial Default Subsets list, then disables the filter by unchecking the box on the RxNorm Output Options tab.

Bug: Unchecking the box does not disable the filter. The RxNorm filter is applied regardless of the state of the checkbox.

Temporary Fix for 2004AB: Do not pick RxNorm filter on the Install menu and then attempt to deselect this option on the RxNorm Output Option tab.

Date identified: 7/20/2004

Date fixed: To be corrected in the 2004AC release.