Cancer Information

so today we're going to be doing ncbi resources for cancer research. i'm peter cooper. as ben mentioned earlier, i'm lead for the education group. and ben is our genomics outreach coordinator. i'm going to be taking you through some of the basic resources at ncbi, focusing on things that are directly linked to the gene database, and then we'll do an example finding literature in molecular data for a gene, cdkn2a. ben's going to talk about some more advanced applications, clinical applications, clinical research, data science, and immunotherapy.

so the ncbi database, one way of sort of breaking them up is to think about three different categories. and one of them is the biomedical literature. pubmed is sort of the premier example of that. the central one and the one that we're going to be focusing a lot on today, are the molecular databases. these include things like dna protein sequences, structures, variations, and chemicals, thinking about pubchem and so forth. then we've recently gotten more involved in some clinical and genetics types of

resources. some of them listed here, gtr, clinvar, medgen, omim, pubmed health, and dbgap. i'm going to be focusing mainly today on the molecular databases, but i'm actually going to do a little bit of the literature, as it is linked to the molecular databases we call genes. and the way we access data at ncbi is really by two main services, the entrez system and the blast system. and entrez is a text search system. it also integrates databases, and you can follow the links to go to other

databases from the one you started with. there are a number of viewers associated with entrez, the graphical sequence viewer, which we'll use today, which is a way of looking at sequences and their annotation; the variation viewer, which is one of our genome browsers. and we'll even get a chance to look at cn3d, which is a viewer that can be associated with structured views on our pages, and it also functions as a standalone application. last is our sequence similarity search tool. it searches all of the dna sequences and protein sequences at ncbi.

there are a number of different services that are listed for you here. i will mention using genome blast data in my talk. we'll also use primer-blast. those are the primer of the exons of a gene. and i think ben is going to do a little demonstration with sra blast during his portion of the talk. so let's start talking about searching at ncbi. this is the entrez system. this is the system you use when you type something into a search box.

there are about 39 different entrez databases there. they're integrated with each other in various ways through links that pass between records. a quick guide to the kinds of data that you can find here, there's literature sequences, expression information, variation information, protein and nucleic acid structures in the structure database. medical genetics, we've already mentioned some of those, and of course pubchem. this is really of subset of the databases.

there's a lot more of them there. for in of them, when you go there you have an interface that looks like this. this is the one for nucleotide. i just wanted to remind you how these things work. some things here are fairly evident. there are filters on the right- and left-hand side that will let you sort of filter down your results to the kind of things that you're interested in. in this case i'm looking at refseq messenger rna. there are various ranging operations you can do on the data there as filters.

and on the right-hand side we have something called the discovery column, and it's going to be different things there, depending on whether you're looking at a search page like this one, or whether you're look at results. this is going to be where you can find things to analyses, perhaps, some that you can run on the fly, some that we've run, or related data of various kinds. you can construct these complex boolean queries. this is not a particularly complicated one. this would be a search that would get you all h1 sequences that are from humans that are not models for example, and i could filter that down and make it

messenger rnas and refseqs if i wanted to. i also have an advanced interface that i won't show you right now, but we'll use it in a few minutes, that will help me construct these kind of queries to get exactly the kind of data i want. one of the big problems facing people using entrez, i think, is this business of the tremendous number of databases and where do you begin if you're looking for a particular kind of data? it's not clear a lot of times. well one way around this problem is to really focus on several of the central

resources, and these are the main entry points to ncbi for a lot of people, and they're good resources for linking out all the other kinds of data that are available. pubmed or taxonomy database, the bio project database that ben's going to talk about in his portion of the talk, the assembly database if you're interested in genome assemblies, and the gene database if you're interested in aspects of a particular gene. and that kind of a query is the kind of query that many, many people have. and i sort of think of the gene as the central access, and that's how i'm going

to use it today to find all kinds of data about a particular human gene. the other thing that's good about a gene is that, you know, you don't have to really write super complex boolean queries to search it. these are real simple things. if you know the gene symbol and you know the organism you can find the record if you're looking for. if you want to download everything, you can download all the gene records for a particular organism. and that current-only filter, by the way, is on by default, so you don't

typically have to include that in your query. moreover, we know that gene searches are pretty important components of what people do in the ncbi entrez system, and so when you type a gene symbol that we recognize in one of these search boxes on the pubmed database or the nucleotide database or the protein database, you're presented with this gene sensor that gives you direct access to information associated with the gene record. and so this is sort of a modified diagram of one that many people at ncbi have on their slide decks, where you have this sort of way of navigating around the entrez system using links and neighbors and things like that.

but these days, really, the place -- the way to do this is not so much that old-fashioned way but to start in some central resource like gene, and then link out to find the things that you need in other databases. when looking for sequences don't search the sequence database. start in gene, and you can usually find the sequences that are associated with that particular gene. okay, so what i want to do now, hopefully with less technical problems, is to jump out of the slides here and to go into a web browser and do some live searches to show you a few things.

so we're going to work with this gene that's called the cyclin-dependent kinase inhibitor 2a, cdk2a, which is actually a complex locus that contains, in a lot of ways, two different genes. we'll talk a little bit about that. we want to look for literature for that gene and its product, sequences. we'll look at the genomic context of that particular gene. we'll look at the structure of that gene in the human genome, and we'll even design some primers for the particular exon gene using primer-blast. then we'll do some other kinds of typical things that people do with molecular

data. i can look at orthologous genes and other species. i can look for some gene expression data. and i'll look at variants, and that's going to sort of set the stage for what ben's going to do. but before i switch over to ben, i'll just do last thing, and i'll show you a 3d structure for this particular approach. and i'm going to start over here on the ncbi homepage. and what i'm going to do is i could search all databases if i wanted to, but i'm

going to show you the gene center, and we'll use that today to do some things. so i'm going to scroll down here and pick pubmed from the pull-down list. and i'm going to search for that gene symbol that i had shown in the slides already, cdk2a. i'll go ahead and search that also, open up my browser site a little bit. so immediately i've done a pubmed search, and it has all the lovely things that pubmed has available. but notice that i'm presented right away with sort of a different alternative to searching pubmed this way, and that's to get information directly from a gene

database. and since i'm in pubmed, one of the things i might want to do is to look at articles that are about gene function. these 2,073 articles are linked to gene records that i have found by doing the search, and those might be a particular relevant set of data -- particularly relevant set of data. notice that it's a smaller set of data than the one i already have. it might be a little more tractable. let's go ahead -- and we're not going to leave pubmed just yet.

let me go ahead and retrieve those articles. so there's 2,073 of them. suppose i'm interested in melanoma. one of the things that i could do is add a term to these records that i got. i can do that by going -- and if i look i can see my search history down here on the bottom. i can go ahead and add one of those terms to my search builder. these are the links that i got from gene. i'll give you my history number up here in the search folder, and then i can add

something to that. one of the important fields is mesh terms. i'll type in melanoma.. this one's wrong. this is the one i want. so i'm indexing this for melanoma in the medical subject headings. i can go ahead and run a search. i could click that filter. and if i want to see ones that are free full text, which is the thing that i'm often doing, of course it's very easy to see that first one there.

now i have a nice review article that i can read the full text up here in pubmed, so cdk2a gene, and another gene, cdk4. so let's go to the gene. now i can get to the gene a couple of different ways. i could click directly on the link to the gene that was right there in the pubmed center, so it's a good way to do it. it could also link here. it saved me having to go back there. and i got the two genes that were mentioned specifically in that abstract.

i'm going to go ahead and click on the cdkn2a gene here to retrieve the gene record. okay, so this is a very interested tumor suppressor gene, and you can read about it here in the summary for this gene. it actually makes two distinct protein products. one of them is called arf, and that's a different kind of suppressor. it interacts with p53. and we'll look at the cdkn2a product today when we're doing our example. so there are several things you can do from here.

one of the first things i want to do is take a look at some sequences. and we actually know that when people come to gene records, this is one of the most visited parts of the record, so let's go down here and take a look at those reference sequences. so this is typical of a human gene. you're going to have reference sequences that exist independently of the build of the genome. there are a couple of different kinds of those. there is something called the refseq gene record, which is intended to be a

genomic standard coordinate system for a particular gene. so that's what that ng refseq is about. and you probably all know that refseqs have particular kinds of accession numbers. then you'll notice that there are a bunch of different splice variants that give rise to different forms of isoforms, here and the one we're going to focus on today is this protein that's often called p16ink4a, and this is an inhibitor of --cyclin-dependent kinase. and notice we have the accession number here for the transcript and for the

protein. i'll just scroll down a little bit further to show you that there is another variant that's interesting, and that's this one. this is p16, and this is another tumor suppressor. this is a totally different protein in a lot of ways, because what happens is there's an upstream exon that we'll see in a few minutes that's part of this transcript. it uses some of the downstream exons but it translates in a different reading frame than the other one, and so it's actually just encroaching altogether.

other sequences that are here, of course, are those that are part of the human genome build, and here is our reference genome here. this gene is on chromosome nine. so let's go up and take a look at some of the structures of this gene. of course i could click through the link to graphical view of chromosome nine. another way to do that is from the center of the gene page here. and you'll see that the graphical sequence viewer is embedded right in the middle of the gene page. we're seeing this region of chromosome nine that has the gene annotated here.

so the tracks that i'm going to focus on today are mainly the gene track and the variation track. so i'm going to get rid of some of these other tracks. i can simply click the xs here to get rid of them. we'll leave the variation tracks in tact. the rna-seq exon covers, i'm going to go ahead and eliminate that one, that one, and that one. so now i'm seeing the cdkn2a gene. you can see there's some complexity associated with the number of difference

kinds of transcripts that it has. i didn't mention this, but i will mention it right while we're looking at these. notice that there are two kinds of transcript accession numbers. the x accessions are the ones that are dependent on that rna-seq data. these are gene models, i would think, that way. and they're considered to be associated with the build itself. so they're not instantiated from pre or standalone sequences in genbank. the n and refseqs are the ones we're going to be focusing on today. those are based on standalone sequences in genbank.

and there is one of interest right there. that's our transcript for the p16. let's zoom out a little bit. we can take a look at the region around cdkn2a. you'll notice that i can see genes that are nearby. notice that there is related -- related cdkn2b. and we'll look for orthologs that are species in a few minutes, and if you look at their genes and in the vicinity, you'll see exactly the same gene structures with these two genes right together.

one of the manipulations you can do on the graphical sequence viewer that's useful is to sort of change the way a particular track is displayed. what i'm going to do is change a gene track so that it shows everything. so i'm going to show -- this is going to show all the protein products and the all the transcripts for a particular gene. it will be helpful for us if we're looking at the proteins for example. i'm going to hit "apply." and so now, to me, this is a little bit nicer display of cdkn2a. notice that if i wanted to zoom into that particular gene, i can mouse over that

object, hit the magnifying glass, and it just zooms into that particular gene. so one of the things i want to point out to you is we're going to be focusing on this product here, the p16ink4a product, and you'll notice that it uses those downstream exons. down here in the variation tracks, which we'll talk more about in a few minutes, you can see that there are a number of variants that are purple. they're purple because they contain some pathogenic variants, as reported in clinvar. we'll talk more about clinvar in a minute.

what i want to do is just zoom in so we can take a look at that. so i can reach up here on the graphical sequence viewer and i can stretch this out, stretch out an area or a region, and i can zoom on that range and zoom into that particular exon. so now you can see quite clearly the number of variants that are associated with this exon and that there's a number of pathogenic variants there. this is a nice piece of sequence that someone might be interested in analyzing. one thing you might want to do is to amplify this, assay something to see if there are variants in there.

notice that you can do that directly from the graphical sequence area in this tools menu. i can do blast and primer search. i could pick this selection and send it directly to primer-blast if i wanted to. but i want to show you another way to do this just because it's a display that's very useful, and that's at the top of the page. or i can change the format. i can change this format to gene table, and i get the graphical sequencer there too, but now i can look at find an exon table for my trance script.

and here it is right here. we can expand that. i can now pick out the exon that i want, and that's that second coding exon, the one that's 307 basis. if i click on that link, what that does for me is it loads into the nucleotide notice that i now can take this piece of sequence and do anything i want with it. i can run blast with it. i could pick primers from it.

so let me just do that real quickly. i can go over here to load chromosome nine with a particular position, and all i need now to do to make this work for me is to pick the appropriate background database, and that's going to be the reference assembly from selected organisms. and 9606 is the tax id for human. i could change that to an english word and it would work just as well. so i can design primers this way. if i wanted to be careful about it, i could even adjust these ranges up here to make sure that i got specific primers that would only amplify the exon by bind

outside of it, so i can easily adjust these ranges. so that's a very handy thing to be able to do from here. i won't run that search for you. there's a handout that goes with this if you want to see that, and you could look in the handout and see you would get very nice results. you can get ten sets of primers that would amplify this particular exon. so it's going to go back to the gene record, so i can change this from gene table back to full report. okay, there's a couple of other tasks that we might be interested in.

one of them is to simply find homologs in our species. and really, what we're talking about from our gene annotation pipelines are orthologs. so those are the corresponding genes, i guess, is a good way of looking at it. so notice there was actually two paralogs together, the n2a and the n2b. we'll be talking about the one that's the correspondent of the n2b. if i click on this ortholog link, these are going to be the ones that were discovered in our genome annotation pipeline. that's a very quick way of finding these.

and at this point they're all going to be from vertebrates. for this particular gene you can look at the organisms here. a good way to see to see what they are is click on this tree link, which sort of tells you what kinds of organisms are here. everything here is a placental mammal. there is one exception. we did identify one bird ortholog, and that's actually the adelie penguin. i'll leave it to you as an exercise to use genomic blast and use the protein sequence for the p16 protein.

and you can easily find that there are homologs in birds like chickens. it's just that they done quite match our criteria for identifying them as all right, let's go back to the gene record. i'll just click through here. that's another kind of task, find these orthologs. another common kind of task is to find gene expression information. you can see some evidence in gene expression if you look at the tracks here that show you the rna-seq alignment. you can also look at microarray data in genome profiles.

and if i follow this, what i'm going to see are these profiles. these are made from geo datasets. they're basically a slice of expression across other samples, in particular, they were applied to a particular microarray experiment, so these are different samples. they could be different treatments of a particular cell line. they could be tumor versus normal tissues, things like that. and so we'll look at the relative expression of whatever particular probe they used to represent this particular gene.

so we have 3,528 of these profiles. in order to use this effectively you'd need to know what experiment you were interested in. one of the things you can do real quick though, is sort of see if there's any differential expression in some of these sets, some of these profiles, and change the sorting order here to sub-group effect. and so these are sort of forming now by the ones where there is the greatest differences between the different samples. a good one is that first one.

these are head and neck tumors. some of them have human papilloma virus infected. i can just click on the profile here to see. and you can see that the hpv-positive tumors tend to have higher level expression of this particular gene. okay. the last, and, probably in many cases, the most interesting thing that we can do is to look at variants associated with a particular gene. i'm not talking about splice variants here, i'm talking about sequence variants,

where there's different changes at the level of the genomic -- the sequence. so if you look, there's a section in the record called "variation," and what i'm going to do, i'm going to load the gene into the variation viewer. we've had a webinar on how to use this particular tool. it's a very handy way of seeing the kinds of variants that you want for a so there's the gene itself. notice i could pick the source database, whether they are in clinvar, what their pathogenicity is like. frequency, so we have the 1000 genome, it's minor [inaudible] that's a new one that's in here now. so let me go ahead and narrow these down.

i'll go to dbsnp. i'll look for variants that are in clinvar. i'll look for pathogenic variants, single nucleotide variants, and i'm entering [inaudible] sequence. these will be the [inaudible] variants here. i now have seven of them. i'll just expand the link for a particular one. this is an interesting one. so the information here in the clinvar information tells me that these variants

are associated with a hereditary cancer syndrome, susceptibility to malignant melanoma for example. notice that this is mapped on all of the transcripts and the isoform of the different proteins for this particular gene, on here in the p16 protein here. and also for arf, a different amino acid chain. so here we've got a methionine changed to iso [inaudible] change. for arf it's going to change to histamine. [inaudible] protein is here. we do have [inaudible] where this methionine isoleucine change is, and that's

structure. so that's easy. i'll go back to the gene record here. in the related information there is a direct link to 3d structures. there are several solution and mr structures of the protein alone. there's also -- this is a kinase inhibitor. this is inhibitor complex in the kinase, in the cyclin-dependent kinase and our tumor suppressor are crystallized together. we can retrieve this record here.

it's actually -- biologically it's a tetramer. just to make life easier, i'll just use the structural asymmetric unit, which is just this dimer right here. and if you look at what's here, the cyclin-dependent kinase 6 is the har this cyclin inhibitor, it's the v chain. and i have our little viewer installed on this computer. if i have it installed i can click on "view structure," and here is our remember, ours is a v chain. it has this anchor here.

and that methionine that i was interested in is up to position 53. i can highlight that for you right there. notice that when i highlight there, it's also highlighted in the structure. and you can see maybe why this might be an important [inaudible]. so that methionine sits right at the junction or the interface of the inhibitor. although the change we're talking is subtle, you can see that it might affect the binding and reduce the effectiveness of this as the tumor suppressor. okay, so now i'm going to wrap that up, and we've basically taken you through lots of different kinds of resources, starting from a single gene record.

and i'm going to turn it over to ben. okay, so i'm going to turn this over to ben busby. hey, everybody, i'm going to talk a little bit about how to use ncbi resources for both clinical research, as well as clinical utility as it pertains to cancer, and then i'll talk a little bit about data science and some of the other educational opportunities we'll have pertaining to data gen viewer if we wanted to look at snp, we would look at just short nucleotide variants. these are going to be under 50 nucleotides. and in this case we can look for things just in the clinvar database.

and i'll talk about in a minute, clinvar is a database of variants that are submitted by clinical sequencing groups and asserted to have some involvement with a particular phenotype. we can select ones that are just pathogenic, and those that, in this particular case, are not short indels but only single nucleotide variants. we can also look for ones that are non-synonymous . i'd like to mention, before we jump over to clinvar, that we can also look at minor ileal frequencies in 1000 genomes, in some cases go esp, which is a large heart study, as well as in the body of 88,000 exomes performed at the grove

institute. if we want to go to clinvar we can take a look at some of these variants and we can see where they are, the transcript change, and the protein change. we see that are presumably more deleterious than others. and also, we can look at the number of independent submitters. in my personal opinion, if multiple [inaudible] submit the same variant with the same phenotype, that is very strong evidence that this variant does have some interplay with that particular phenotype for a whole bunch of phenotypes as we open -- as we expand and close the different variants.

so now i'll switch back to the presentation and i'll talk about one particular variant, just as an example. so this is a particular variant that peter cooper was discussing earlier. the one thing i'd like to point out is that in the general research use consortium of dbgap, we have looked at these variants and seen how many individuals in dbgap have these particular variants. this particular variant is extremely rare in that dbgap general research use consent group, but it may very well be in other studies in dbgap. another thing, i like genetic testing registry.

so these are clinically improved tests, in most cases, for particular genes or for particular conditions. in fact, on the gtr homepage you can search for tests by either genes, a cdkn2a, we can see that there are 79 tests which cover that gene for a variety of different conditions. going back to clinvar, once again, we can see with this variant that we've called this particular variant. so there are three parts to talk about quickly. one is our new advanced search interface.

also, i'd like to mention that you can download data dictionaries, scientific variables, although not individual level data for those particular phenotypic variables from the public ftp site, and there's a link right of the front of the dbgap homepage. finally, we have a phenotype/genotype integrator where you can search gwas studies by either and find genotypes by gwas. so over 50,000 in the 63 studies have been diagnosed with cancer, and their data has been deposited in dbgap. typically these individuals also have match controls.

if we use the advanced search page we can search for melanoma, for example, very quickly, and we can find, in just a second or two, that there are seven studies with patients with melanoma, many of them with many patients, and we can see that some of them employ snp chips or snp genotypes and others use next-generation sequencing. in the case of the second study here, whole genome sequencing, to look at these things. what i will show you a little bit later on in this presentation is how to deal with that next-generation sequencing data.

in fact, we'll talk about two ways to deal with different formats in the next-generation sequencing data. once again, for some particular studies, you can look at analysis and you can see p values for particular chromosomal locations derived from gwas. if you wanted to do a larger scale search in gwas, you could use our phegeni tool and search for a phenotypic trait, such as melanoma, or you could search for a gene, such as cdkn2a. here i've searched for melanoma, and i can see that there are several gwas studies, which have unearthed a number of intergenic and intronic snps related

to melanoma. i should note that the data is the same as that coming from the nhgri gwas catalog. we can also view these particular snps on a diagram of the human genome if that's something desirable. going back to dbgap for a minute, doubtless a few of you belong to large genomic sequencing consortia. if you do, the data produced by your consortia can be back-ended by dbgap and still presented in a front-end sort of way by [inaudible].

please e-mail me at ben -- b-e-n -- .busby -- b-u-s-b-y -- @nih.gov. finally, i'd like to point out that the general research use collection now has 71 studies and counting. so studies in this collection can be used for a number of applications that are not necessarily disease specific, such as answering genomic structure per se, which is something many cancer researchers are interested in. now i'd like to point out that the whole genome sequencing data or exonic sequencing data that is in dbgap is in vertical databases, and that's stored in what we call sra format.

now the sra database itself contains two kinds of data; first, that which is controlled access or in dbgap, as well as public access. so as it pertains to cancer, or specifically melanoma in this case, we would be talking about mostly cell-line data. also, we can see that we have both dna and rna data, as well as a bit of metagenomic data for melanoma, which is becoming more and more interesting to cancer researchers. and then it's also important to note that some of this data is aligned and some of this data is not.

and as i'll show you in a few minutes, that's how we most easily investigate that particular data. excuse me. however, one of the earliest things that investigators typically like to do when they're presented with a list of datasets is to investigate the metadata associated with those datasets. at ncbi, in my personal opinion, the easiest way to do that is to click on the "send to" menu in the top right of your screen and then select "run selector." this will send up to 10,000 runs that you have searched for to the run selector.

currently, i'm including both aligned rna data that is public access pertaining here i can see that, actually, this is just align data for melanoma, and what i can see is that i have a whole bunch of columns of metadata. now please remember that this crosses a number of studies, and not all of the studies will have provided us all of the metadata columns. that said, if you are providing data and associated metadata to ncbi, please give us as much metadata as possible, as it is really crucial for users to be able to use your data. here i've shown you a number of the columns of datasets.

here, if you scroll down the right, you see many, many, many columns. another thing i'd like to point out is that here you can see a whole bunch of runs -- given the bio project by simply clicking on the bio project number in the center of our screen. the nice thing about this is that it will give us pre-analyzed data, which is in geo, all of the biological metadata associated with the sample, which is also in the run selector in the bio sample database patients that are associated with the study, and in my personal opinion, hopefully pmc. another really neat thing about using the sra data structures is that now gatk

and hisat work directly with data that is deposited in sra, either in [audio difficulties] i have sent one such sra experiment to blast, and here i'm blasting in, the gene that peter cooper talked about several times in the beginning of this presentation; namely, nm_000077.4. one thing i did before blasting this in to the sra was that i removed the poly a tail from this dataset. as you can imagine, i'm blasting into millions of raw reads or soft mask reads from next-generation sequencing experiments, and it is ideal to cut off the poly

a tail, because one will get many hits to the poly a tail. i encourage you to try this, as well as similar blast searches, at home. before we wrap up, i'd like to spend just a couple of minutes informing you about some other resources we have that are pertinent to immunology, but particularly to immunotherapy, which is now a very popular both experimental and therapeutic avenue for some cancers. i'm not going to go in detail into any of these databases, but if you have additional questions, please put them in the question pod or e-mail us at help at help@ncbi.nlm.nih.gov.

so the first thing that i'd like to talk about briefly is clone db. so in the past many, many immunological sites have been sequenced by cloning, and so we have maintained those resources, and you can search for them here. another resource that we have that many people are not aware of is something called "dbmhc." here we have a database of very common mhcs, and there's a variety of a tools associated with that database. also, we have igblast capability. i've seen a number of talented bioinformaticians write scripts, essentially to replicate this functionality.

you should know that this is available for you, both as a web resource, or if you're a bioinformatician, as standalone blast. we've given several webinars on standalone blast and would be willing to give some more. finally, i did not talk about structural variants, but obviously structural variants are extremely important when we think about immunology and immunotherapy. dbvar is our database of structural variants, which are typically nucleotide variants over 50 nucleotides.

obviously some structural variants can go up to megabases, and one can search dbvar to look for things like those pertaining to t-cell receptor. i've also filtered these for those that have been submitted to clinvar as pathogenic, and as you can see, there are a very large amount. here, i'm showing some examples, and as you can see, these particular examples range from a couple hundred nucleotides all the way up to several mega bases. finally, as many of you know, the geo is an extremely useful resource, but one thing many people don't know is there is quite a lot of epigenomic and methylation data in the geo resource, particularly pertaining to immunology and

amino therapy. if you search geo or bio project you can filter by these types of data. another thing many people don't know is that we have many bed files we support, and those can also be found in the geo database, or in bio project, and they can be viewed on our sequence viewers. with that, i would simply like to point out some resources that we have that you can go to learn more. if there is a specific resource that you would like to see a webinar or a short ncbi minute on, please don't hesitate to contact us.

and if you have general questions, please e-mail info@ncbi.nlm.nih.gov if you'd like to see particular webinars, please e-mail webinars@ncbi.nlm.nih.gov and if you need help with blast, of course, e-mail blasthelp@ncbi.nlm.nih.gov i hope you've enjoyed the webinar, and thank you very much for your time. hopefully you've had a great day.