Searching TSCA Spectraless Database Using NIST Search Jan. 07, 2013

Transcription

1 Searching TSCA Spectraless Database Using NIST Search Jan. 07, 2013 Before using the CAS Registry and SciFinder, we searched the TSCA database and our PM (Plant Material) list using the NIST search (Poster Sessions, ASMS 2001, 2004). We no longer update copies of the TSCA database in NIST format. The "spectraless" database contained names, CAS Number, molecular weights, and molecular formulae. However, others might still find the capability useful so I have included information on using the approach. We purchased TSCA database on CD and converted it into NIST binary format for searching. Originally we could only search by molecular formula and nominal molecular weight and other various constraints, but the current NIST software allows the TSCA database to be searched by monoisotopic mass with an error window. The main problem with the approach was that the structures were only shown for compounds which were present in the NIST database. The structures in the NIST EI database were correlated to TSCA entries by corresponding CAS numbers. Thus, the absence of structures for many TSCA entries made data review tedious in many cases. Futhermore, there was no way to prioritize the list by the number of associated references or key words as we can do we CAS and ChemSpider searches. Current changes in the NIST search software also allows user-created databases such as TSCA to also be searched by monoisotopic mass with an error window. This might be useful if there is some doubt in determining a molecular formula from the accurate mass data. To access the menu, Select the TSCA library by itself from the menu,

2 Add the Search value, specify mass, and an uncertainty..

3 Then select OK from bottom of above menu. You should get the following results for this example.

4 If there are no results returned, one might need to (Re)Index Exact Mass.

5 This reindexing creates a file in the TSCA NIST library folder named exactmw.inu. In this document, I have included some changes in the approach, a handout for the ASMS poster session, and a copy of the poster session.

6 Conversion of File on Solutions, Software Corp. TSCA CD (May 2004) They have changed the distribution format of the files on their CD. First find the Tsca xls file (date attached will change every time). Open it and save as a Text (tab delimited file) to the C:\tsca folder which contains the tsca.exe program.

7 Then rename the tsca txt to tscainv.dat. Then run the tsca.exe. It should convert the tscainv.dat to a tsca.hpj file in a few minutes. A log file, tsca.txt will be created with information on the records converted. The following log info was noted in the current conversion. TSCA DATA RUN ON 10/28/2003 Total exported: Total polymers: Total Complex: Total unspecified MF: Total examined: 64978

8 The only problem with this approach is that some name records will have appended to the end of their name. Somehow this occurs when the.xls file is saved to a tab delimited file which puts the undesired before and after the name. Fortunately, the tsca.exe program is not bothered by the before the name. Examples of a record field and its conversion are noted below. The following record is from the orginal tsca xls file: RN: IN: Guanidine, monohydrochloride SN: DF: FN: ; ; ; ; ; ; ; MF: CH5N3.ClH UV: FL: The IN field is changed to the following format in the Text (tab delimited) file created from Save As command: RN: "IN: Guanidine, monohydrochloride" SN: DF: FN: ; ; ; ; ; ; ; MF: CH5N3.ClH UV: FL: The IN field is then converted into the ##CAS Name field in the final tsca.hpj file with the residual noted at the end of the field: ##TITLE=Library Entry 1 in TSCA ##JCAMPDX=Revision 4.10 ##DATA TYPE=MASS SPECTRUM ##SAMPLE DESCRIPTION= TSCA Listing ##NAMES= ##CAS NAME=Guanidine, monohydrochloride" The tsca.hpj file is converted into a NIST library binary folder and associated files using the Lib2Niste.exe utility program supplied by NIST. The libraries thus created are searchable by MW, CAS number, MF, etc using the NIST library search software.

9 Identifying Unknowns with MS Data Using TSCA, In-House, and CAS Registry Spectraless Databases Introduction Mass spectrometrists routinely employ electron impact (EI) and computer-search programs to identify unknowns with commercial databases (NIST, Wiley, etc.). Even though theses databases have dramatically increased in size and diversity, there are still many commercial organic compounds of commercial interest absent from the libraries. For example, we estimate that the NIST98 database of 107,886 entries only contains 8042 out of ~64,000 components in the TSCA Inventory. Granted, many of these TSCA entries cannot be characterized by electron impact mass spectrometry. However, we estimate that ~39,000 can be characterized by mass spectrometry when including newer soft ionization methods. These newer methods include electrospray, thermospray, field desorption, FAB, chemical ionization, and MALDI. Unfortunately, identification of unknowns with data from these methods is not as routine as computer-searching EI databases. This paper describes methods that we employ to identify unknowns by searching spectraless databases employing LC/MS, LC/MS/MS, and in-source LC/MS fragmenation data. These databases contain no spectra (EI, MS/MS, etc), thus we refer to them as spectraless. Databases searched include the TSCA (Toxic Substances Control Act) Inventory, the Chemical Abstract Services (CAS) Registry File, and our in-house database of plant materials (PM). LC/MS/MS and in-source LC/MS fragmenation data are added to user libraries in NIST format. These spectra can then be used to help identify unknowns by searching the database by spectrum or by structure. Software and Procedures to Prepare Molecular Weight Correlated Database Creating TSCA Database in NIST Binary Format: The TSCA Inventory is obtained in ASCII format on a CDROM biannually from the following company: SOLUTIONS Software Corporation, a division of RegScan, Inc. 800 West 4th Street Williamsport, PA USA Tel: (570) x1415 or x1105 Fax: (570) Web: solution@env-sol.com Order TSCA-S for annual subscription of 2 issues $93.00 We primarily search the TSCA database using the NIST (National Institute of Standards and Technology). The TSCA inventory on the CD is first converted in a Hewlett-Packard (HP, Agilent) JCAMP ASCII format. This ASCII format is then converted into a binary NIST database that is searched with either Version 1.7 or 2.0 of the NIST library search software. This approach is very convenient since the molecular weights are not present in the original TSCA file. The NIST library utility automatically calculates and includes the molecular weight in the NIST binary file. The CD includes Adobe Acrobat Software for searching the database by molecular formula, but we find the formula search within the NIST library search software more convenient. Search by molecular formula in the Adobe Search field employs formula entered in the Hill Order. Formulas written in Hill Order list the carbon atoms first, then the hydrogen atoms, followed by the remaining atoms in alphabetical order. If carbon atoms are not present in the formula, all atoms are entered in alphabetical order. Conversion of TSCA Database into JCAMP Format. Fred Wampler at Eastman Chemical Company wrote the conversion program. The source code for the Pascal program is shown in Figure 2. The program, tsca.exe (stand-alone program, no runtime program required), is run on a personal PC with either Windows

10 9X, Windows NT, and Windows We have not tested it with Windows Millenium Operating System. It filters out characters in the original TSCAINV.TAB file, which could not be displayed (those NOT between chr(32)...chr(126)). There were many such characters in the original file. The program was originally written to convert the TSCAINV.DAT file. However, currently the file is name TSCAINV.TAB on the CD. Therefore, TSCAINV.TAB must be renamed to TSCAINV.DAT when copied from the CD to the PC s hard drive. The file TSCAINV.DAT is placed in a directory with the program tsca.exe. It creates a file named tsca.hpj and a text file named tsca.txt. The tsca.txt contains information on the conversion. Contents of the tsca.txt file from a recent conversion are shown below: TSCA DATA RUN ON 9/23/2000 Total exported: Total polymers: Total Complex: Total unspecified MF: Total examined: The TSCA file contained entries, but many of the entries are mixtures, polymers, or contain no molecular formulae. Only were thought to be useful for searching by molecular weight by parameters included in our conversion program. The conversion program tsca.exe filters out the unwanted entries. The complexes include salts and mixtures. The molecular formulae of the complexes or salts are separated by a period. In general, the most useful molecular weight information was found to the left of the first period in a complex. Therefore, only these parts of the molecular formulae fields in the TSCAINV.TAB file were exported to the tsca.hpj file for entries. A typical record in the TSCAINV.TAB file and its converted form in the TSCA.HPJ file are shown in Figure 1. Many commercial mass spectrometry library search software packages require that a spectrum be included for each entry. Thus, the electron impact spectrum of hydrogen was entered for all records. In our conversion program, the molecular weight field is not calculated. The calculation is not needed because the NIST Lib2Nist.exe (see next paragraph) automatically calculates the nominal molecular weight during the creation of the NIST binary database, TSCA. The nominal molecular weight is determined by adding the molecular weights of the most abundant isotopes. For example, the molecular weight of methylene chloride (CH2Cl2) is , and thus its nominal mass is 84 in the NIST database. Often organic chemists calculate the molecular weight ( Chemist s molecular weight ) using the weighted average of the masses of the elements. For example, the Chemist s molecular weight of methylene chloride is This latter molecular weight was not used for the molecular weight field in the NIST database. The molecular formulae of salts are not as one would expect. This makes it somewhat difficult to search for unknowns unless some chemical information indicates their presence. Several examples are shown to demonstrate this point. For example, 1,3-benzenedisulfonic acid disodium salt s molecular formula (CAS No ) is listed as follows:

11 SO - 3 Na+ SO - 3 Na+ Listed as: C 6 H 6 S 2 O 6. 2Na Not: C 6 H 4 S 2 O 6. 2Na or C 6 H 4 Na 2 S 2 O 6 The listing of sulfonic acid salts and other salts in the TSCAINV.TAB file whose cation is known are all consistent with the example. The anion and cation are split by a period and the cation is replaced with an equivalent number of protons. In other words, the molecular formula to the left of the period is for the free acid. This is very convenient for our needs since we normally obtain the molecular ion -hydrogen for organic acids in negative ion mode (normally do not see doubly charged ion by electrospray, FAB, LSIMS). The molecular weight field in the NIST binary database was created by summing only the molecular formula to the left of the first period in molecular formula field in the TSCAINV.TAB file. Therefore, the molecular weight used to search the database for compounds that are acids should be converted to their free-acid form before searching the database (see example problem listed in later section). N,N,N-trimethylmethanaminium bromide s molecular formula is (CAS No ) listed as follows: + N Br - Listed as: C 4 H 12 N. Br Not: C 4 H 12 BrN In tetra-alkylammonium salts such as the example above, the molecular formula field in TSCAINV.TAB file lists the cation to the left of the period and the anion to the right of the period. The molecular weight field in the OPUS database was created by summing only the molecular formula left of the first period. Therefore, the molecular weight used to search the database for tetra-alkylammonium salts should employ the molecular weight of the cation (charged species). The molecular weight or molecular formula of the unknown from mass spectral data is used to search the TSCA inventory list. The TSCA inventory does not include a molecular weight field. Thus we convert a commercial version of the database available on CD-ROM into a JCAMP formatted file. The JCAMP file is converted with a NIST utility, Lib2Nist.exe, into a NIST binary format. The TSCA database can then be searched by using the NIST Search Program (Version 1.7 or 2). The NIST98 databases of 107,886 chemicals only contain 8042 chemicals found in the TSCA database. Thus our TSCA database is very useful for identifying chemicals in commercial products. The number, 8042, was determined by selecting the mainlib database in the NIST search and employing the other search (Sequential; constraints; MW range 1-10,000; Other databases TSCA).

12 There are many other compounds in the TSCA Inventory that are filtered out in the formation of our NIST database. Surfactants are a good example of chemicals that are not included in our binary database that are easily analyzed by LC-MS. A surfactant example is shown: RN: IN: Hexanedioic acid, polymer with.alpha.-hydro-.omega.-hydroxypoly(oxy-1,2-ethanediyl) SN: Adipic acid, polyethylene glycol polymer DF: FN: MF: (C6H10O4.(C2H4O)nH2O)x UV: FL: XU Another program would need to be written to compile the surfactants and some program to search the oligomeric mixtures by repeat units. Conversion of TSCA JCAMP File to NIST Binary Format: The tsca.hpj file is in the appropriate format needed for creating a database for searching with the NIST Search software (Version 1.7 or 2.0). The tsca.hpj was converted from ACSII to NIST binary format with the NIST library utility, Lib2Nist.exe. The Lib2Nist user-interface with the options button selected and appropriate options selected is shown below:

13 The tsca.hpj file could be converted to other binary formats for searching with other commercial mass spectrometry programs. We tried importing the file into a database that could be searched by the HP Chemstation software, but the presence of such elements as deuterium and tritium in the molecular formulae of some entries precluded this approach. Creating Plant Material Database in NIST Binary Format: A very similar approach is employed to create a NIST database of our approximately 6,000 chemicals that are either produced at or purchased for our manufacturing sites. The PM database is maintained on a server for other uses. It is converted into JCAMP file employing a user-written executable. Information on Searching the Databases NIST Search of TSCA and PM Spectraless Databases: We regularly create updated TSCA and PM databases in a NIST binary format. We can then easily search the databases using standard functions found within the NIST Search (Version 1.7 or 2.0). Fields that are searched include nominal molecular weight, molecular weight range, molecular formula, CAS number, name fragment, elements present, and elements value. The software also includes a sequential search, which allows multiple constraint fields to be searched. Search CAS Spectraless Registry File: A wider variety of chemicals can be accessed by searching the American Chemical Society s Chemical Abstract Service (CAS) Registry File. CAS Registry includes compounds found in chemical journals; chemical patents; TSCA; and other chemical lists from Canada, Korea, and Europe. It is estimated that the CAS registry includes >15 million entries containing >9 million as discrete compound which could possibly be characterized by either LC-MS, solid probe mass spectrometry, or GC-MS. We normally employ SciFinder for searching the CAS Registry by molecular formula. Accurate mass data must be used to calculate a discrete molecular formula for searching, since the CAS Registry cannot be searched by nominal mass with SciFinder. SciFinder is a PC application which is very simple to use. Better searches can be performed (e.g. crossing molecular formulae or molecular weight with substructure, name fragments, etc.) can be performed with STN Express from American Chemical Society s Chemical Abstract Service. STN Express is somewhat difficult to use. Thus, we usually get someone from our Technical Information to assist us in these searches. The molecular weights for components are calculated with the values shown in Appendix 1. These molecular weights are calculated differently than those in the NIST Search Software! Conclusions Spectraless databases such as TSCA, our in-house plant material, and the CAS Registry are very useful for the identification of unknowns. These approaches are especially important when employing soft ionization data obtained from electrospray or APCI LC-MS. The TSCA Chemical Inventory is searched with molecular formulae or molecular weight data. Substructural information obtained from the insource fragmentation data is then used to narrow the candidate list. The TSCA database is obtained on CD and converted into a binary database, which is searched with the NIST Search software. The CAS Registry File is searched with SciFinder or STN Express.

14 Figure 1: A typical record in the TSCAINV.TAB file and its converted form in the TSCA.HPJ File TSCAINV.TAB Record: RN: IN: Formaldehyde SN: Formaldehyde DF: FN: ; ; ; MF: CH2O UV: FL: TSCA.HPJ Corresponding Record: ##TITLE=Library Entry 1 in TSCA ##JCAMPDX=Revision 4.10 ##DATA TYPE=MASS SPECTRUM ##SAMPLE DESCRIPTION= TSCA Listing ##NAMES= ##CAS NAME=Formaldehyde ##MOLFORM=CH2O ##CAS REGISTRY NO= ##MP= e+02 ##BP= e+02 ##MW= ##$RETENTION INDEX= e+00 ##$CONDENSED SPECTRUM=NO ##NPOINTS=1 ##XYDATA=(XY..XY)

15 Figure 2: Delphi/Pascal Code Used for Converting From TSCAINV.TAB Format to TSCA.HPJ Format unit tscaunit; interface uses Windows, Messages, SysUtils, Classes, Graphics, Controls, Forms, Dialogs, ExtCtrls, f_gen; type TForm1 = class(tform) Panel1: TPanel; private { Private declarations } public { Public declarations } end; var Form1: TForm1; implementation {$R *.DFM} var infile,outfile,statfile : textfile; buffin: array[ ] of char; buffout: array[ ] of char; rn,v_in,mf,nline : string; cnt,cntp,cntun,cntplx,cntnull,cntexp,cntempty :integer; start_rn,start_in,start_sn,start_mf,start_uv : integer; // Input file text is terminated not with a linefeed/carriage return but // with a carriage return only. procedure f_readln(var fhand:textfile;var outline:string); const // Keep printable characters goodchar: set of char = [#32..#126]; var schar : char; begin outline:=''; repeat read(fhand,schar); if schar in goodchar then outline:=outline+schar; until schar = #10; end; begin // make sure input file is available if not fileexists('tscainv.dat') then exit; assignfile(infile,'tscainv.dat'); filemode :=0; system.settextbuf(infile,buffin); reset(infile); assignfile(outfile,'tsca.hpj'); rewrite(outfile); system.settextbuf(outfile, buffout); assignfile(statfile,'tsca.txt'); rewrite(statfile); cnt:=0; cntp:=0; cntun:=0; cntplx:=0; cntnull:=0; cntexp:=0; cntempty:=0; while not eof(infile) do

16 begin // read input file line by line and process based on tags //(MF;RN;IN;SN;UV) etc. f_readln(infile,nline); inc(cnt); if pos('mf: (',nline)>0 then begin // count various types inc(cntp); continue; end; // ignore Unspecified if pos('unspecified',nline)>0 then begin inc(cntun); continue; end; // calculate needed positions within the line start_rn:=pos('rn:',nline)+4; start_in:=pos('in:',nline)+4; start_sn:=pos('sn:',nline)+4; start_mf:=pos('mf:',nline)+4; start_uv:=pos('uv:',nline)+4; rn:=copy(nline,start_rn,start_in-start_rn-4); v_in:=copy(nline,start_in,start_sn-start_in-4); mf:=copy(nline,start_mf,start_uv-start_mf-4); if length(trim(mf))=0 then begin inc(cntempty); continue; end; inc(cntexp); // keep info to left of period if pos('.',mf)>0 then begin mf:=copy(mf,1,pos('.',mf)-1); inc(cntplx); end; // write to output file writeln(outfile,'##title=library Entry 1 in TSCA'); writeln(outfile,'##jcampdx=revision 4.10'); writeln(outfile,'##data TYPE=MASS SPECTRUM'); writeln(outfile,'##sample DESCRIPTION= TSCA Listing'); writeln(outfile,'##names='); writeln(outfile,'##cas NAME=',v_in); writeln(outfile,'##molform=',mf); writeln(outfile,'##cas REGISTRY NO=',rn); writeln(outfile,'##mp= e+02'); writeln(outfile,'##bp= e+02'); writeln(outfile,'##mw='); writeln(outfile,'##$retention INDEX= e+00'); writeln(outfile,'##$condensed SPECTRUM=NO'); writeln(outfile,'##npoints=1'); writeln(outfile,'##xydata=(xy..xy)'); writeln(outfile,' '); end; // write to status file writeln(statfile,'tsca DATA RUN ON ',datetostr(date)); writeln(statfile,' Total exported: ',inttostr(cntexp)); writeln(statfile,' Total polymers: ',inttostr(cntp)); writeln(statfile,' Total Complex: ',inttostr(cntplx)); writeln(statfile,'total unspecified MF: ',inttostr(cntun)); writeln(statfile,' Total examined: ',cnt);

17 // close open files system.closefile(infile); system.closefile(outfile); system.closefile(statfile); exit; end.

18

19

20 (Banner) Identifying Unknowns with LC-MS Data Using the TSCA Chemical Inventory or SciFinder James Little, Eastman Chemical Company, Kingsport, TN

21 1 Overview Identify unknowns in competitive products, environmental samples, etc. with molecular weight/molecular formula from MS/MS or in-source data, EI, CI, etc. Spectraless Databases searched such as TSCA, in-house plant materials, CAS Registry File TSCA and in-house plant material databases searched with NIST Library Search Program CAS Registry File searched with SciFinder or STN Express

22 2 Introduction The TSCA (Toxic Substance Control Act) Chemical Inventory, CAS (Chemical Abstracts Service) Registry, and our in-house listing of plant materials are very useful for the identification of unknowns. These approaches are especially useful when employing soft ionization methods such as electrospray or APCI LC-MS. The databases are searched with either molecular formulae or molecular weight data. Substructural information, obtained from MS/MS or in-source fragmentation data, is used to narrow the candidate list. Examples using these approaches to identify unknowns and the mechanism for converting the commercially available TSCA database into a NIST-searchable database are discussed.

23 3 TSCA Database TSCA purchased on CD twice per year 2. Converted to JCAMP ASCII Format (~40K entries) 3. Converted to NIST Binary Format for Searching (Calculates Nominal MW During Conversion from MF)

24 4 Steps in Identifying Unknowns with TSCA Database -Either MW, or even better, MF data obtained with TOF LC-MS -MS/MS or in-source data obtained for substructural Information - Search TSCA Database Searched Using MW or MF and substructural information -Add MS/MS or in-source to user library for identifying similar compounds in future

25 5 In-Source Fragmentation Essentially Same as MS/MS If Components Resolved by LC N 2 V Increase V, Increase Fragmentation MH + N2 V Fragment Ions

26 6 In-Source Fragmentation Spectra of Unknown 25 v 50 v 75 v MH three times, (CH 3 ) 2 CH=CH 2 from t-butyl groups or CH 3 CH=CO from propanoate groups?

27 7 MW Search Results, One Consistent with Fragmentation Determined from Name

28 8 MF Search Results, Only One Candidate

29 9 75 Volt In-Source Spectrum Interpreted Irganox 1019 MH x 56 loss 1 x 56 loss +2H s m/z 335 m/z 318 m/z m/z x 56 loss 4 x 56 loss

30 10 Other Useful Search Constraints within NIST Version 2 Search Primarily Search MF and MW, but others options include: -Element value (Cl = 1, Cl = 1-3, S < 2, etc.) -List of elements considered -Name fragment -Sequential (mixture more than one constraint) Also structural searches of components added to MS/MS and in-source user databases

31 11 SciFinder and STN Express Searches Of CAS Registry File SciFinder: -Only searchable by MF data, 19 Found for Irganox example 2 consistent with fragmentation, see Poster 12 -Easy to learn and use PC interface STN Express: -Searchable by MW (nominal) and MF -More difficult to use than SciFinder, usually get help from our Technical Center -MW crossed with other constraints to get reasonable # of candidates

32 12 Two Possible Structures from SciFinder Search

33 13 PM (Plant Material) Database PM Database on Corporate Server; purchased and manufactured chemicals 2. Converted to JCAMP ASCII Format (~7K entries) 3. Converted to NIST Binary Format for Searching (Calculates Nominal MW During Conversion from MF)

34 14 Conclusions Spectraless Databases such as TSCA, CAS Registry, Plant Material very useful in identifying unknowns in samples using MF, MW, Substructural information, name fragments, etc. TSCA and Plant Material Databases Searched with NIST Ver 2 MS Search Software CAS Registry Files Searched with SciFinder and STN Express Information available from jllittle@eastman.com or directly at

35 15 Other Information on Web Site -Selecting CI gases, Building CI Manifold, Library Search Info, Accurate Mass, Silylation and Diazomethane, Doubly Charged Ions, Sailing, etc. at Acknowledgements -Steve Stein and Yuri Mirokhin, NIST -David Sparkman, Consultant for NIST -Bill Tindall, Kent Morrill, Fred Wampler, Eastman Chemical Company