[BioC] allen brain atlas
Radek Blatny
blatny at img.cas.cz
Mon Aug 25 16:01:11 CEST 2008
Hi Martin, thank you very much for the additional code. Not having a
batch query is strange since the web interface has a possibility to
query multiple genes at once (separated by coma) so I would expect to
have this possibility in the API as well. Another solution might be
to download the whole expression data in an xml file and parse it
into an R table-type object. However I wasn't able to find the whole
db in an xml file.
Regards, Radek
On Aug 25, 2008, at 3:47 PM, Martin Morgan wrote:
Hi Radek --
"Radek Blatny" <Radek.Blatny at img.cas.cz> writes:
> Hi Martin,
>
> the code works perfectly, thank you! I've got one more question. Is
> there
> a way to do a batch query?
>
> When I tried to send a list of symbols to the script, the process
> stopped
> with an error if one of the genes was missing in the ABA and there
> was no
> data retrieved at all. Any suggestions?
It seems like the ABA lets you query individual terms, or to do a
search, but not to do a batch-style query. A solution might be
abaBatchGenes <- function(symbols) {
res <- lapply(symbols, function(symbol) {
tryCatch(abaGene(symbol),
error=function(err) {
list(gene_expressions=data.frame(
avgdensity=numeric(),
avglevel=numeric()),
image_series=data.frame(
imageseriesdisplayname=character(),
age=integer(),
imageseriesid=integer(),
sex=factor(levels=c("F", "M"))))
})
})
names(res) <- symbols
res
}
and then
> abaBatchGenes(c("Coch", "Foo"))
> Ideally, I would like to send a list of gene symbols, skip the genes,
> which are not present in the ABA and retrieve a dataframe
> containing the
> records consisiting of a gene symbol and expression values for each
> brain
> region.
Backing up, and with your specific objective in mind, one might
abaGene1 <- function(symbol) {
url <- abaUrl("gene", symbol)
xml <- xmlTreeParse(url, useInternal=TRUE)
data.frame(symbol=symbol,
structurename=xq(xml, "//structurename"),
avgdensity=as.numeric(xq(xml, "//avgdensity")),
avglevel=as.numeric(xq(xml, "//avglevel")),
row.names=seq_len(xq(xml, "count(//structurename)")))
}
abaBatchGene1 <- function(symbols) {
res <- lapply(symbols, function(symbol) {
tryCatch(abaGene1(symbol),
error=function(err) FALSE)
})
do.call("rbind", Filter(is.data.frame, res))
}
and then
> abaBatchGene1(c("Coch", "Foo", "Calb1"))
failed to load HTTP resource
symbol structurename avgdensity avglevel
1 Coch Cerebellum 7.888339 11.977408
2 Coch Cerebral cortex 52.086498 46.967388
[SNIP]
16 Coch Striatum ventral region 6.137324 4.482482
17 Coch Thalamus 27.540497 36.667355
18 Calb1 Cerebellum 67.864638 95.345704
19 Calb1 Cerebral cortex 76.402445 89.302411
[SNIP]
33 Calb1 Striatum ventral region 88.071941 87.418639
34 Calb1 Thalamus 79.896777 87.737941
Martin
> Thanks, Radek
>
>> Hi Radek --
>>
>> Radek Blatny <blatny at img.cas.cz> writes:
>>
>>> Hi, is there a way to access the expression data from the Allen
>>> Brain
>>> Atlas and use them for annotation of a topTable? I am particularly
>>> interested in being able to download expression profiles for gene
>>> symbols (or other suitable identifiers), which is possible to query
>>> in semiquantitative graphical format for all brain anatomic regions
>>> at the ABA webpage - for instance here:
>>>
>>> <http://mouse.brain-map.org/brain/Myst4.html?ispopup=1>
>>
>> Not sure of a package, but it might be easy enough to create
>> something
>> useful yourself. Here's a quite fun first attempt:
>>
>> library(XML)
>>
>> ## some utitlity functions
>>
>> abaUrl <- function(type, id = "") {
>> if (!missing(id))
>> id <- paste("/",id, ".xml", sep="")
>> paste("http://www.brain-map.org/aba/api/", type, id, sep="")
>> }
>>
>> xq <- function(xml, query) {
>> unlist(xpathApply(xml, query, xmlValue))
>> }
>>
>> ## retrieve information about a gene; separate some of the info into
>> ## two data frames
>>
>> abaGene <- function(symbol) {
>> url <- abaUrl("gene", symbol)
>> xml <- xmlTreeParse(url, useInternal=TRUE)
>> res <- list(gene_expressions=data.frame(
>> structurename=xq(xml, "//structurename"),
>> avgdensity=as.numeric(xq(xml, "//avgdensity")),
>> avglevel=as.numeric(xq(xml, "//avglevel")),
>> row.names=1),
>> image_series=data.frame(
>> imageseriesdisplayname=xq(
>> xml, "//imageseriesdisplayname"),
>> age=as.integer(xq(xml, "//age")),
>> imageseriesid=as.integer(xq(
>> xml, "//imageseriesid")),
>> sex=factor(xq(xml, "//sex"))))
>> free(xml)
>> res
>> }
>>
>>
>> With this one can
>>
>>> coch <- abaGene("Coch")
>>> coch
>> $gene_expressions
>> avgdensity avglevel
>> Cerebellum 7.888339 11.977408
>> Cerebral cortex 52.086498 46.967388
>> Hippocampal region 2.939453 5.846055
>> Hippocampal formation 4.755450 6.896857
>> Hypothalamus 8.690369 9.506371
>> Lateral septal complex 8.783522 7.551568
>> Midbrain 9.267348 11.813993
>> Medulla 7.031233 12.667056
>> Olfactory bulb 26.625715 34.037346
>> Pons 3.897534 5.504243
>> Pallidum 8.595852 9.827189
>> Retrohippocampal region 7.256083 7.999830
>> Striatum-like amygdalar nuclei 7.874489 9.268417
>> Striatum 74.447227 79.294968
>> Striatum dorsal region 100.000000 100.000000
>> Striatum ventral region 6.137324 4.482482
>> Thalamus 27.540497 36.667355
>>
>> $image_series
>> imageseriesdisplayname age imageseriesid sex
>> 1 Coch-Sagittal-06-0609 55 75990683 M
>> 2 Coch-Coronal-05-2779 55 71717614 M
>>
>>
>> For more fun I used the EBImage package and a little more exploration
>>
>> ## Retrieve an image
>>
>> library(EBImage)
>>
>> abaImageSeries <- function(imageseriesid) {
>> url <- abaUrl("imageseries", imageseriesid)
>> xml <- xmlTreeParse(url, useInternal=TRUE)
>> res <- list(images=data.frame(
>> imagedisplayname=xq(xml, "//imagedisplayname"),
>> downloadImagePath=xq(xml, "//downloadImagePath"),
>> stringsAsFactors=FALSE))
>> free(xml)
>> res
>> }
>>
>> abaImage <- function(downloadImagePath, zoom=1) {
>> url <- paste(abaUrl("image"),
>> "?zoom=", zoom,
>> "&path=", downloadImagePath, sep="")
>> readImage(url)
>> }
>>
>> and then
>>
>>
>>> series <- abaImageSeries(coch$image_series[1,"imageseriesid"])
>>> series
>> $images
>> imagedisplayname
>> 1 Coch_2
>> 2 Coch_10
>> 3 Coch_18
>> 4 Coch_26
>> 5 Coch_34
>> 6 Coch_42
>> 7 Coch_50
>> 8 Coch_58
>> 9 Coch_74
>> 10 Coch_82
>> 11 Coch_90
>> 12 Coch_98
>> 13 Coch_106
>> 14 Coch_114
>> 15 Coch_122
>> 16 Coch_130
>> 17 Coch_138
>> 18 Coch_146
>> 19 Coch_154
>>
>> downloadImagePath
>> 1
>> production18/Coch_06-0609_38068/zoomify/primary/0203016726/
>> Coch_2_0203016726_A.aff
>> 2
>> production18/Coch_06-0609_38068/zoomify/primary/0203016726/
>> Coch_10_0203016726_B.aff
>> 3
>> production18/Coch_06-0609_38068/zoomify/primary/0203016726/
>> Coch_18_0203016726_C.aff
>> 4
>> production18/Coch_06-0609_38068/zoomify/primary/0203016726/
>> Coch_26_0203016726_D.aff
>> 5
>> production18/Coch_06-0609_38068/zoomify/primary/0203026750/
>> Coch_34_0203026750_A.aff
>> 6
>> production18/Coch_06-0609_38068/zoomify/primary/0203026750/
>> Coch_42_0203026750_B.aff
>> 7
>> production18/Coch_06-0609_38068/zoomify/primary/0203026750/
>> Coch_50_0203026750_C.aff
>> 8
>> production18/Coch_06-0609_38068/zoomify/primary/0203026750/
>> Coch_58_0203026750_D.aff
>> 9
>> production18/Coch_06-0609_38068/zoomify/primary/0203034952/
>> Coch_74_0203034952_B.aff
>> 10
>> production18/Coch_06-0609_38068/zoomify/primary/0203034952/
>> Coch_82_0203034952_C.aff
>> 11
>> production18/Coch_06-0609_38068/zoomify/primary/0203034952/
>> Coch_90_0203034952_D.aff
>> 12
>> production18/Coch_06-0609_38068/zoomify/primary/0203044879/
>> Coch_98_0203044879_A.aff
>> 13
>> production18/Coch_06-0609_38068/zoomify/primary/0203044879/
>> Coch_106_0203044879_B.aff
>> 14
>> production18/Coch_06-0609_38068/zoomify/primary/0203044879/
>> Coch_114_0203044879_C.aff
>> 15
>> production18/Coch_06-0609_38068/zoomify/primary/0203044879/
>> Coch_122_0203044879_D.aff
>> 16
>> production18/Coch_06-0609_38068/zoomify/primary/0203054806/
>> Coch_130_0203054806_A.aff
>> 17
>> production18/Coch_06-0609_38068/zoomify/primary/0203054806/
>> Coch_138_0203054806_B.aff
>> 18
>> production18/Coch_06-0609_38068/zoomify/primary/0203054806/
>> Coch_146_0203054806_C.aff
>> 19
>> production18/Coch_06-0609_38068/zoomify/primary/0203054806/
>> Coch_154_0203054806_D.aff
>>
>>> img <- abaImage(series$images[1,"downloadImagePath"], 2)
>>> img
>>
>> 'Image'
>> colorMode() : Grayscale
>> storage class : numeric 3D array, writable images in range [0..1]
>> dim() : 513x414
>> fileName() : /tmp/magick-XXxlP9B8
>> compression() : JPEG
>> resolution() : dx = 72.0, dy = 72.0
>>
>> image 1/1:
>> [,1] [,2] [,3] [,4] [,5] [,6]
>> [1,] 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922
>> [2,] 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922
>> [3,] 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922
>> [4,] 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922
>> [5,] 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922 0.9803922
>> ...
>>> display(img)
>>
>> This displays an image at a sort of reasonable resolution; it's then
>> possible to use EBImage to do all kinds of fun manipulations. There's
>> an incredible amount of scope for programmatic manipulation here.
>>
>> The way to construct the queries above, and the structure of the
>> return data, are described on the 'API' page, e.g., following the
>> link
>> to 'API documetation' on this page:
>>
>> http://community.brain-map.org/confluence/display/DataAPI/Home
>>
>> Martin
>>
>>> Regards, Radek
>>>
>>>
>>> Radek Blatny, MSc.
>>> Institute of Molecular Genetics
>>> Department of Mouse Molecular Genetics (C/O Jiri Forejt)
>>> Czech Academy of Sciences
>>> Videnska 1083
>>> 142 20, Prague
>>> Czech Republic
>>> Tel. (+420) 241 062 260
>>> Fax (+420) 241 062 154
>>> http://www.img.cas.cz/mmg
>>> email: blatny at img.cas.cz
>>> Skype name: blatny
>>>
>>> _______________________________________________
>>> Bioconductor mailing list
>>> Bioconductor at stat.math.ethz.ch
>>> https://stat.ethz.ch/mailman/listinfo/bioconductor
>>> Search the archives:
>>> http://news.gmane.org/gmane.science.biology.informatics.conductor
>>
>> --
>> Martin Morgan
>> Computational Biology / Fred Hutchinson Cancer Research Center
>> 1100 Fairview Ave. N.
>> PO Box 19024 Seattle, WA 98109
>>
>> Location: Arnold Building M2 B169
>> Phone: (206) 667-2793
>>
>
--
Martin Morgan
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109
Location: Arnold Building M2 B169
Phone: (206) 667-2793
More information about the Bioconductor
mailing list