12.07.2015 Views

View - ResearchGate

View - ResearchGate

View - ResearchGate

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

164 Osborne et al.amount of filtering and is useful when accuracy is desired; the “moderate” modelis similar to the strict model but lacks syntactic filtering and is best suited to evaluateinput text as a whole rather than as discrete phrases. Finally, the relaxedmodel provides minimal filtering of its component strings and is best used forexploring. In cases wherein accuracy is important, select a strict data model.When doing exploratory fishing for associations, it is suggested to start with arelaxed data model and move to a moderate data model if the results are undesirable.If one is going to use a different data model one will have to download itfrom the download page mentioned in the Subheading 3.3.1.3.6.2. The Perils of Filtering With MMTxFiltering of data is an option with MMTx, but often it can cause more problemsthan it solves. However, filtering may remove undesirable matches, it alsohides the fact that such matches occurred. High-scoring matches may get pastone’s filter, but one will not know how to remove them if one has no rankinginformation. By fully mapping the text it is possible to programmaticallyremove high-scoring but low-ranking matches that are counterproductive to thedata mining at hand. For the same reason it is cautioned against removingsources from contention unless it is for licensing purposes or the source causesmore problems than it creates for one’s mappings.3.6.3. Running From the Command Line (for Nonprogrammers)The details of running MMTx are found at http://mmtx.nlm.nih.gov/runMMTx.shtml and two examples are provided with MMTx usage detailsshown on a separate webpage at http://mmtx.nlm.nih.gov/semanticTypes.shtml.The actual text is the fifth column, which is of interest in mapping. The othercolumns could be eliminated by using a spreadsheet or the UNIX “cut” option,but in this case one can use the MMTx input parameters to handle fielded text.One will also select the option “show_cuis” which is turned off by default. Thisallows to actually determine if one has mappings to the concepts of interestwithout having to manually investigate the text. One can also turn off the candidatesand mappings (-c=false and -m=false) to reduce the amount of output anduse the sections option to specify the entire line from which geneRIF wasderived. The new version of MMTx (MMTx 2.4B) takes a semantic type as anargument (⎯restrict_to_sts=neop) not yet specified in the documentation. Byspecifying the abbreviation for neoplastic process one can restrict one’s resultsappropriately. The actual command to run MMTx appears as follows:MMTx ⎯fieldedText ⎯textField=5 ⎯fieldSeparator=’|’ ⎯fileName=generifs_basic⎯show_cuis -c=false -m=false –-sections ⎯restrict_to_sts=neop > outputfile.txtIn this command the input file is specified (—fileName=generifs_basic) infieldText format (⎯fieldedText), separated by the “|” character (⎯field

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!