 So, what we're going to be doing, and it's a shorter one, because this will give you time to work on the assignments, we're going to be looking at databases. I've talked a little bit about some of them. I've introduced you to a couple briefly, but I think it's important to understand a few more of them and be aware of some of them. We'll look at some of the other databases that aren't really just chemical databases, but are really intended to be NMR or MS databases, or pathway databases as well, and then there's another group of databases that we call comprehensive metabolomic databases. So, one of the things that's been going on for a long time is that bioinformatics has been in one silo, and another field called cheminformatics has been in another silo. And they emerged as different fields of information technology at different times. Cheminformatics actually began in the 1970s. It was a very development of chemistry-based software, chemical drawing software, chemical databases, American Chemical Society, CAS databases, and things like that. This was developed before the internet. It was developed before widespread use of personal computers. They were developed before sort of the ubiquity of computing. And they were also developed at a time when software was considered, and should only be considered as a commercial product. So the result is that most cheminformatics tools software, most of the chemical software that was developed was, and largely still is, commercial. Bioinformatics developed in the 80s, really, and it emerged with the growth of the internet, and the model of free, open-source, open-access software. And a challenge has been as we realize that there's some useful chemistry stuff out there, and we need chemistry to do biochemistry and biology, that trying to get those tools and databases from the cheminformatics world and make them open-access, open-source, and free has been difficult. And to some extent, you actually have to build all of our own stuff all over again using the bioinformatics model. So as I said, this is the history, as I say, of the cheminformatics 60s and 70s, user pay, limited access, a lot of it was produced by companies. Bioinformatics through the 80s and 90s was designed from like your biologists, web-based, grew with the internet, and it was funded by public agencies with the expectation of public access. A lot of the software that we use depends on databases. Most of the bioinformatics software now is fundamentally dependent on GenBank, PDB, Uniprode, and Geo, and ArrayExpress. So what databases do is they allow you to consolidate and link information. They allow you to do query or information retrieval. They typically contain a lot of reference values, or reference data, or reference sequences, like protein or DNA sequences, or reference images, chemical structures. A lot of people use databases as well for training, and this is an important area because a lot of what we want to do is prediction in order to be able to predict something you have to practice. So if you have a training set, you can practice on these databases of training set, and then you can use them to test. A key utility as well in databases is similarity searching. So most people in the world of bioinformatics know about sequence similarity searching, but you can also do text similarity searching, structural similarity searching, which is also used in chemistry, spectral similarity searching, and image similarity searching on Google Images. And then the other aspect of databases is to help with aspects of prediction. The primary utility of GenBank is actually to predict gene function, and the annotations that people put into GenBank and the species associations, those are the critical things. Sequences are just tags, hashtags, to allow you to get the function information. So people have used Protein Data Bank to predict structures. They're very good at it now. We can predict properties from structures and from databases, again partly through learning. We've learned about phylogeny, activity, and other relationships. So prediction and similarity searching are some of the main applications for databases in both chem informatics and bioinformatics. Databases evolve, and in the world of metabolomics, it's a young field, and so many cases, many of the databases that are out there are what we call hobby databases. And this is actually how GenBank started. It started as a hobby that a fellow named Russell Doolittle had, and he was collecting sequences in the literature and just typing them in at night. Another person, Margaret Dayhoff, who is considered the mother of bioinformatics, she also started databases as a hobby, and she published books on them in the 1960s, and these are just Protein Sequence databases. Anyways, they were all flat files. Dayhoff databases weren't even flat files, they were books. Doolittle's database was a flat file. But then they actually started doing some things with the databases. As the database grew, Russell Doolittle found that if you started doing a sequence search against viral genes, you could actually identify oncogenes. And the first identification of an oncogene is what allowed sequence searching to sort of take off. As the database grows, and as people want to add more data to it, eventually you start creating what are called curated databases. So this is again an example of SwissProat, which is a predecessor to Uniprot, was a curated database that was handled exclusively by people at the Swiss Bioinformatics Institute. They would choose sequences that they liked. They would curate them and add to them. You couldn't deposit sequences into them. The database grew and grew and grew. And eventually it got actually so large that they had to decide to either go bankrupt or seek permanent government funding, which they eventually got through the EBI and it became Uniprot. As those databases grow and the importance grows, then eventually they can become archival databases. And this is what GenBank and the Protein Databank are. As yet there is no truly archived open deposition database for metabolomics. There's a database called Metabolites, which is disemerging. But it's, I think, only about 30 people have deposited data into it. So it's more a hobby database at this stage in terms of its size. But as these grow, the community grows. The user community grows and the greater demands about it. They don't have to go from the limited coverage that a hobby database may have to very extensive coverage that more archival relational distributed databases have. It also costs a lot more. As the database gets larger and more complex, it becomes very, very expensive. And I can say that a huge amount of MyLabs resources go into maintaining the databases that we offer. It's largely unsustainable. So we're not sure how much longer we'll be able to offer these databases. Likewise, as databases grow, you have to have increasing needs both for standardization. And this is what the Metabolomic Standards Initiative or MSI is about. Better ways for exchanging and sharing information, standard markup language formats, ZML, NMRML, much greater dependence on annotation and automation. You can't do everything by hand, whether it's filling in dates, associating species or descriptions. You can't do it with teams of people after a while. You actually have to depend on computers and text mining. Likewise, as things grow, people want to be able to query in more and varied ways. And so the need for relational databases becomes greater. And web interfaces, graphical visual interfaces. I've seen this slide before, whether it's genomics and proteomics and the sort of missing database and missing killer app that hasn't been around, which allows you to take raw spectra or raw data and interpret it in terms of concentrations and metabolite identifiers. So, up until recently, most metabolomic data was really in textbooks. And it goes back a long, long ways. There's a lot of chemical information at 75, even 100 years old. Arguably the whole field likes behind proteomics and genomics by about 20 years if we look at the database development and when first gene and protein sequence databases first appeared. The other thing is that, in fact, the diversity of users in metabolomics is actually far greater than it is for proteomics and metabolomics. You're looking at not just metabolomics researchers, but analytical chemists are interested in plants and clinical chemists, physicians, drug researchers, people in the field of spectroscopy, people in bioinformatics, people who are involved in standardization and standard formulation. So that's another challenge. And this is one of the things that's maybe why metabolomics has been lagging behind some of the other fields. So there are, I guess it could say, five major classes of databases, the NMR databases, the mass spectral databases, the chemical or compound databases, pathway databases, and then ones that combine essentially all of the above and we'll call those comprehensive metabolomic databases. Usually the comprehensive databases are metabolomic specific, but there are others that are fairly generic. And many of them are aware of at least some of these and probably use some of them every day. So the NMR spectral databases, this is a list of some of them, a spectral database, SBDS, the NMR shift DB is a chemical shift database for NMR, organic molecules mostly, metabolomics, medicine metabolomics, consortium database, and the biomegres bank. The STBS is a database maintained in Japan and it's sort of like the NIST of Japan, it's called AIST, and they collect data and you are able to access and query both NMR, they have IR and UV data, they have thousands of compounds, but you're not able to freely download their data, they have very strict limits on what you can access. So it's been around for a long time, lots of mass spectra, although a lot of that mass spectra data was collected on very old mass spectrometers so it's not quite relevant. Lots of NMR spectra, thousands, lots of FTIR and UV spectra, they have very nice search tools. As with the NIST database and for GCMS, most of the compounds are not metabolites though. As large as it is, compelling as it is, it still isn't something that would relate frequently to the metabolomics community. Biomegres bank, I mentioned this briefly, but this historically was a database for depositing NMR data for large molecules, probably proteins, but about five years ago it shifted into using depositing data for small molecules and they've been steadily adding to a database of more than a thousand molecules where they have NMR spectra and the websites have proved considerably over the last couple years in terms of annotation and in fact the most heavily used part of the biomegres bank ironically is its metabolomic component. So it's now more than 800, it's closer to 1,000 now. They've collected about five or six NMR spectra for each compound. You can search by chemical names, formulas, smiles, strings. A lot of metabolites in this were planned, it was done as a rabidopsis effort, but they've started throwing in mammalian metabolites so it's a pretty broad collection of compounds. A lot of them are now assigned so you have data that not only gives you the spectra but also the assignments. NMR ShiftDB is a database that was developed by Chris Steinbeck who now heads up the metabololites section at the EVI, he also directs Kebby. This database has about 50,000 NMR spectra. It's totally open access, totally downloadable. In that regard it's probably better than the SBDB but again most of the compounds are non-metabolics. These are organic synthetic compounds, most of them dissolve in chloroform or TMSO so not in the typical water or aqueous conditions that we use. You can look at some of the structures there and you can see most of them are not terribly biological. But it is a collection of chemical shifts and structures that are potentially useful. It does have a tool for chemical shift prediction. Given that some of these tools are expensive, it's nice to know that there's at least one that's free. Now it's easy to search by names and structures and chemical shifts. As I say, most of those are in organic solvents so that doesn't make it as relevant to metabolomics. The Metabolome, Madison Metabolomics Consortium Database is really an extension of the Biomagras Bank. So they just basically took all the Biomagras Bank data and put it into another database. But they started adding literature, chemical shifts, literature data so they archived that. So it has an additional set of literature chemical shifts that you wouldn't find. It also includes some mass spec literature derived spectrum. So through that you can do structure name searches, chemical shift searches, and then it includes additional data about the compounds, formulas, names, properties, and images, and then links to other databases. Now mass spectral databases, this is essentially an identical slide to what I showed you guys before when we were talking about mass spec. So I won't dwell on this very much, but the major pure mass spec databases are Metlin, NIST, GOM, and Mass Bank. How many people have used Mass Bank or ever seen Mass Bank? One. Anyways, it's a really extensive collection and in fact the group that runs it really is hoping that it will be used more and more by metabolomics. So it's being developed specifically for metabolomics in some respects. Although the data that it's got is sort of scattered from a wide variety of areas. Has anyone heard of the database called Respect? It's another mass spec database, but it's focused primarily on polyphenols and polyphenol and phytochemical metabolites. So Mass Bank has peak searches. It's actually a very good interface for doing peak searches. It's a fairly intuitive structure. It's very well maintained. It has a wide variety of spectra. It's collected on multiple platforms, triple quads, TOS, FTMS, INTRAPS, QTOS. It covers about 15,000 compounds and collects data from many, many countries. So in that regard, again, it's probably a step up over what you might find even in Metlin. So I talked about Metlin before. We talked about GOM, so I just wanted to highlight Mass Bank. Chemical compound DBs. We've seen these as well. We've talked about Kebby. We've talked about PubChem and ChemSpider. There's also another database called Ligand Expo. So Kebby, as I said, is about 30,000 high quality compounds. They grab them a lot from other databases from KEG, from LipidMaps, from Drug Bank. The focus in Kebby isn't about spectra. They're not about chemicals or biology or their biological interpretation. It's about their ontology, classifying and naming compounds, which seems that it's a little odd in terms of the emphasis because most of us aren't that interested in ontologies of chemicals, but that's what it was established to do. PubChem, again, it's upwards of 30, 35 million compounds, 35 million substances. The requirement is that rather than 1,500 Dalton's, they typically want something that's less than maybe 10,000 Dalton's, so it includes some peptides, large peptides. They get their data from other people. They really don't generate their own. The data is deposited by vendors, and they worked out agreements to have the vendors deposit their data. So Sigma, Aldrich, the Zinc database and others. So there's quite a bit of redundancy, actually. That's why they said 75 million substances. It's about only 32 million unique compounds. PubChem has grown from just being a pure chemical database. They started to add more and more information and linkage to PubChem, to improve their search, to improve the quality of their annotations as well. PubChem-Spider was a duplicated effort somewhat, and it's sort of been a second fiddle to PubChem for a long time. It's maintained in the UK, largely, or by people in the UK. It's a little bit distinct from PubChem because they actually try and include a bit more data. They're very, very focused on the synonym sets, and the synonym sets in PubChem-Spider generally are better than what you get in PubChem. They also are trying to link more and more to pharmacology, to spectral data, so they're trying to add value that you might not find in PubChem. So sometimes there's information in PubChem-Spider that you don't find in PubChem and vice versa. Lincoln Expo is a database, actually. It's of interest because it contains the small molecules that are bound to the proteins in the protein data bank. So as we noted at the beginning, lots of small molecules play an important role as cofactors and activators and signaling molecules for enzymes and proteins. But what it does is it helps link chemicals, metabolites, and drugs to their drug, or metabolite targets. And that's this important linkage between the metabolome and the proteome and to function. And I think people really underutilize this resource. Lincoln Expo gives you three-dimensional structure, not just the two-dimensional structure of a chemical. So that also is important because the three-dimensional structure ultimately is what determines the biological function of a small molecule. So you can search it through names and identifier codes and formulas and smile strings and inchy and other things. There are more specialized compound databases. There's one called 3DMet, so the 3D structure of metabolites. Napsack, which is a plant metabolite database. Zinc, which is a collection of compounds used for drug screening. And then LipidMet, which is a very large collection of lipids that was put together by LipidMet Consortium. So 3DMet has about 8,500 natural compounds in it, so it's mostly KIG. Napsack, about 50,000 plant metabolites, but those are linked to species information. So that's what makes it actually quite valuable. Fortunately, I don't think there's a whole lot of somewhat limited structure data. Zinc is a database of commercially available compounds. This is mostly a drug bank. And then LipidMet has about 30,000 compounds covering most of the major classes of lipids. Originally, LipidMet was supposed to be about lipids from humans. What they ended up doing actually was downloading a huge amount of lipid data that was maintained by another Japanese database, which covered lipids from everything, from plants, microbes. And so what LipidMet has is the same data in this Japanese database, which covers all kinds of non-human lipids. So it's a little spotty in that regard. So those are compound databases, specialized in general. We've looked at some of the mass spec databases. We've talked about some of the other NMR databases. And so there's the pathway databases. And perhaps everyone's favorite database in metabolomics is KIG. There are other pathway databases, Preactome, which is maintained here, I guess, in Toronto. And it's linked to EBI. Bio-psych, Meta-psych, Ecosych. Those are the psych databases that there's several hundred now, I think, maintained by Peter Karp. And then another one that probably none of you have heard of called the Small Molecule Pathway Database. So what the pathway databases do is they start linking, just like LigandExpo links metabolites to proteins. These pathway databases link metabolites to genes, to proteins, to diseases, to signaling events and processes. So that's critical. That's the biological information interpretation that we're all interested in. And we'll talk about these pathway databases and interpretation more tomorrow when we talk about how to go from these lists to biological interpretation. They have a number of tools, interfaces to align to visualize, map genes and metabolites. Many of them cover multiple species. Which is both beneficial but sometimes confusing for some people. If I say keg, it's something that everyone knows or uses and it allows you to visualize pathways. You can see the EC numbers, enzyme classification codes. You can click on things and it's going to point to brief summaries of compounds and their name. About 10 to 15 data fields per compound. So it's enough to get you at least partially familiar with what's there. Right now keg has about 17,000 compounds in it. It also has nine, almost 10,000 drugs. The drugs, they sort of double, well quadruple count. So every drug, it has a salt form. So aspirin with sodium, aspirin with potassium, aspirin with other types of salts. Each of those is considered a different drug. Which as I say is sort of a cheap way of giving lots of compounds. Very extensive glycan database. Very unique, hardly used. There's over 400 pathways in keg. So that's quite extensive. But remember that covers pathways for many different classes of organisms. So for humans, there's only about 80 pathways. And other mammals, about 80 pathways. And you get into microbes, perhaps 150, 200, you get into plants. You get quite a few more. And so when all of those different pathways for the different kingdoms of life are added together, you get about 400 distinct pathways. Small molecule pathways, database, SmithDB. It's a different kind of database that's specifically designed around humans. Or if you want mammalian systems. But really it's about human metabolism and human pathways. Pathways, there's actually more pathways in SmithDB than in keg. It also covers pathways that aren't in keg. Drug pathways, none of those are in keg. A lot of disease pathways, none of those are in keg. And then signaling pathways, again, none of them in keg. And then there's about 10 metabolic pathways that aren't in keg and another 80 that are in keg. So it covers a component of pathways that you don't standardly get in keg. You can search and browse something that's not so easy in keg and it also allows you to do metabolite mapping. The other thing that's important is that metabolism is not about just simply a wiring diagram without context. Metabolism actually takes place in very specific organs and specific organelles and that information is not deliberately not in keg. And that's unfortunate because it's often critical to understanding metabolism. It's important to know the way that metabolism is happening in the mitochondria or the proxiesome outside the membrane, inside the membrane. Whether it's primarily the liver or the muscles in the brain. All of those things are vital to understanding both the function and the role. And so that's what's attempted to be captured in SMIPDB. So this is a pathway diagram in SMIPDB. So in a sense it's a little different than a keg diagram. You can see that proteins are green blobs. The cofactors are marked in pink. You can see various organelles, proxiesome it looks like. You can see the cell membrane. You can see the organ where this particular pathway happens to be. Structures are not just dots. They're actually drawn into the map so you can actually see the chemical structures. The wiring diagrams aren't all straight lines. If there are cycles and events to be seen cycles are drawn. Things are being transported in and out of the cell. Those are drawn in and out of the mitochondria. Those are drawn. And everything is linked. Metabolites are linked to HMDB and proteins are linked to uniprot. And there's also ways for you to highlight on the right side by clicking checkboxes. Certain metabolites so you can visualize. And then each pathway actually has a description of what's involved and what happens. Which again is something you don't get in KIG. Here's a pathway that happens in the mitochondria. You can see the mitochondria sitting inside the cell and which molecules are being used. In this case you can use HMDB. You can type in a list of metabolites that are increased or decreased, reduced. And it will produce pathways or links to the pathways that have been altered. And again you can use checkboxes to highlight things and then take screenshots or data dumps of those pathways. You can map concentrations as well to metabolites in pathways. If you've measured absolute relative concentrations you can enter that and see where things are, where there's a bottleneck, where there's increase or reduction along the pathway. The last database set I guess we'll talk about are the comprehensive metabolite databases. And we've heard about some of them already. We've heard about the human metabolism database. We've talked about KIG. We've heard about MMCD and also drug bank. So these are databases that do or have everything that all the other databases have. So they have or at least some component. So they have either NMR or MS data. They have pathway data. They support searches. They support everything that I've talked about in these other tools. Now some of them have everything. Some of them are missing a few things. Another set of metabolite databases that again most people may not be aware of are the human metabolism databases about human or mammalian metabolism. There's a yeast metabolism database which is about yeast metabolites and the E. coli metabolism database which is about E. coli. Each of these databases has about 2,000 metabolites and has concentration data of intracellular and extracellular for these cells grown on different media. It has very extensive pathways, connections and other pieces of information. These databases correct a lot of errors that are in the psych databases which have many compounds that are not found in yeast or E. coli. In terms of the HMDB, the human metabolism database, we've talked about this before, there's about 40,000 metabolites, both endogenous and exogenous. At 120, we know of being bacterial metabolites. There's probably many more, but it's hard to tell because many bacterial metabolites look just identical to our eyes. Bacteria produce alanine, we produce alanine. Bacteria produce glucose, we consume and produce glucose. So we can't tell. But the 120 unique bacterial metabolites, things like trimethylamine and HPHPA, other compounds are unique and only produced by microflora. Database has lots of concentration data, both on the normal and abnormal, absolute concentrations. Data on hundreds of diseases, thousands of NMR spectra, MS spectra and GCMS spectra. You can search the database using sequences, you can search the database using spectra, you can search the database using text. You can search by pathways, you can search by structure, by drawing structures. And you can download, essentially, all of the data. It grew out of a project that was launched about seven, eight years ago at the University of Alberta. Some people were involved, particularly Jeff at the very beginning, I think, at the very beginning of the project. It was a very small-scale version of the human genome project. Human genome project was financed for $3.4 billion. Human metabolism project was financed for $7.5 million. Human genome project had 15 years to complete itself. We were given three years to complete ours. Anyways, we didn't want to start from scratch. There's a lot of data that was out there. So part of the effort was to start compiling the data from what we could find in the literature. So we developed a lot of text mining tools. We also do our own experiments and confirm and validate that. All of the data that we've done in the human metabolism project is freely available. That includes the human metabolism database, drug bank, food DB, food components, toxin, toxin target database, or T3DB, and then YMDB and ECMDB and a few others. We also were mandated to develop a lot of technology and software tools. Many of you please will be using in the next few days to help with the field of metabolomics. You guys have seen this slide before, but this, as I say, these represent the information that we've compiled through that human metabolism project, both through experimental work and literature mining and collecting data from many, many sources. So these things are housed in these different databases. The most popular database actually is Drug Bank. It gets almost twice as many hits, like seven million hits a year. And the reason why it's so popular is it was the first database to link drugs to the drug targets. And so all of the major pharma industry used Drug Bank because this is how a lot of them are using it to find new drug targets and to repurpose drugs. All the biotech industry was using this. And a lot of pharmacists and medical professionals and others make use of it. It's not a really large database in the sense that it's only covering about 1,500 compounds. But the linkage to drugs and drug targets is what people find most useful. So Drug Bank was the first database we made, and then we moved to applying the same concepts to human metabolism. So one of the features about the human metabolism database is it links all metabolites to their enzymes and to their targets and to their receptors and to their transporters. So it's not just a list of compounds, like Pubcam. It's actually a list of biology, reaction, pathways, mechanisms, linking metabolites to genes, to proteins, to function. Same thing was done with T3DB, which is a toxin target database. So that's to deal with herbicides, the pesticides, and other sides that we all end up eating inadvertently, or that end up in our blood, just from being exposed to industrial pollutants. It's a small database of about 3,000 compounds. There are other databases that are much, much larger in terms of toxins, and all of those are compounds that have never left the lab. So that's, again, an issue. It seems to plague metabolomics. Yes, you can find a goldmine of data of compounds, but they're mostly irrelevant because they have never left a synthetic lab, and therefore they can never be in any organism. The last one is FoodDB, and it's sort of marked in gray because it hasn't been publicly released, but it's a very extensive database on food components, food additives, but also some of their health effects and their functions and their origins in different plant and food species. HMDB contains a lot of information, as I said. It includes pathways, it includes spectral information, it includes details on structures and descriptions of what the compounds are, where they're found, what their concentrations are, what's normal and abnormal, their disease associations and so on. You can browse it. There's lots and lots of data fields, so they're really more like encyclopedic databases. We've gone through some of the spectral searching tools with HMDB. You can try them in the next exercises, perhaps. Pathways, so everything in HMDB that's linked to SMIPDB is in there, so you can view and search by metabolites and pathways. You can look through different biofluids like urine or cerebral spinal fluid or plasma and serum, or about 15 different biofluids. And it's being used quite a bit by clinical chemists because the data is somewhat unique. Drugbank, as I said, is another resource and it was actually developed before HMDB, partly as a trial run. This structure is very similar in terms of offerings and the type of information that it has. Some of it's more specialized to what drug researchers and medicinal chemists want to know as opposed to metabolomics researchers. As a way of browsing, you can search by chemicals the same way you can in HMDB, so you can draw structures, search by sequences the same way you can do in HMDB. You can also extract data through what we call the data extractor, so it's a MySQL query system, but it's user-friendly. You don't have to know MySQL. This is a summary slide just to compare what you might find between these different types of databases. There are spectral databases, there are chemical compound databases, there are pathway databases, and there are comprehensive databases. What I've marked in red are things that you look for like nomenclatures, synonyms, links and references, presence of spectra, presence of pathways, presence of structures, presence of descriptions, chemical properties, and physiological data. Some of the databases have some of this data, some of it have a little bit of it, some more than others, some have partial data, which I've marked in yellow, so it varies. Depending on your applications and your specialization and interest, some of these databases are going to be more useful than others. There are strengths and weaknesses in each of them. I've certainly been plugging the HMDB, but it's not a database that'll cover plants, and it's not a database that covers yeast or E. coli or fruit flies. Keg does cover that. It is a multi-species database, but it doesn't have the mass back, it doesn't have the NMR, it doesn't have descriptions. Without that kind of context, it makes it a little limited for metabolomics. In the future, I think we'd like to see databases that are perhaps as diverse as Keg, but as comprehensive as HMDB. That's going to take a long time, but it's slowly happening. We'll see where it takes us. What we're going to do now is for the next hour or until you peter out until nine o'clock. We have some exercises, and Michelle brought that up. We're going to have you look at you can either continue working with economics, some of you might find that compelling or interesting, or want to refine that. There's an exercise handout that was given to you in the book. There's also a tutorial that we have. And some of you are just listering along. You can try and do all three. There are other options as well, to look at Metabo Analyst. The key thing is that the reason why you have your computers here and why you're connected to the web, all of these tools require that. So I would really encourage you to try and do things. Simply listening to me maybe gives you some context, but all you need to learn is to by doing things. So we've charged you up with drinks and coffee and fruit if you need more to keep awake. Use that. But hopefully you can spend the next hour and part of the evening tonight to see if you can finish out those things. And Jeff and I will be here for a bit to help. I'm pretty much worn out from this long day.