[CookBook] BioC annotation package

라벨: Informatics
이메일로 전송BlogThis!X에 공유Facebook에서 공유

작성

© 최광민 2011-04-06

제목

블로그 운영원칙

순서

1. Using org.Hs.eg.db
2. Using biomaRt 
3. Making xref db

# Using org.Hs.eg.db

columns(org.Hs.eg.db)

 [1] "ENTREZID"     "PFAM"         "IPI"          "PROSITE"    
 [5] "ACCNUM"       "ALIAS"        "CHR"          "CHRLOC"     
 [9] "CHRLOCEND"    "ENZYME"       "MAP"          "PATH"       
[13] "PMID"         "REFSEQ"       "SYMBOL"       "UNIGENE"    
[17] "ENSEMBL"      "ENSEMBLPROT"  "ENSEMBLTRANS" "GENENAME"   
[21] "UNIPROT"      "GO"           "EVIDENCE"     "ONTOLOGY"   
[25] "GOALL"        "EVIDENCEALL"  "ONTOLOGYALL"  "OMIM"       
[29] "UCSCKG"    

select(org.Hs.eg.db, keys="AB371497", keytype="ACCNUM", columns=c("PFAM", "SYMBOL"))

    ACCNUM PFAM    SYMBOL
1 AB371497 <NA> USP30-AS1


select(org.Hs.eg.db, keys= "SNX13",keytype = "SYMBOL", columns=c("REFSEQ", "ENSEMBL", "PFAM", "SYMBOL")) 

    SYMBOL       REFSEQ         ENSEMBL    PFAM
1   SNX13    NM_015132 ENSG00000071189 PF02194
2   SNX13    NM_015132 ENSG00000071189 PF00615
3   SNX13    NM_015132 ENSG00000071189 PF08628
4   SNX13    NM_015132 ENSG00000071189 PF00787
5   SNX13    NP_055947 ENSG00000071189 PF02194
6   SNX13    NP_055947 ENSG00000071189 PF00615
7   SNX13    NP_055947 ENSG00000071189 PF08628
8   SNX13    NP_055947 ENSG00000071189 PF00787
9   SNX13 XM_005249672 ENSG00000071189 PF02194
10  SNX13 XM_005249672 ENSG00000071189 PF00615
11  SNX13 XM_005249672 ENSG00000071189 PF08628
12  SNX13 XM_005249672 ENSG00000071189 PF00787


# Using biomaRt

library("biomaRt")
 
###
### gene start
###
ensembl <- useMart("ensembl", dataset="hsapiens_gene_ensembl")
getBM(attributes=c('refseq_mrna','chromosome_name','start_position','end_position','strand'),
filters = 'refseq_mrna', values = 'NM_033453', mart = ensembl)

refseq_mrna chromosome_name start_position
1 NM_033453 20 3189514
end_position strand
1 3204516 1

### 

### transcript_start

###
library("biomaRt")
ensembl <- useMart("ensembl", dataset="hsapiens_gene_ensembl")
getBM(attributes=c('refseq_mrna','chromosome_name','transcript_start_position','transcript_end_position','strand'),
filters = 'refseq_mrna', values = 'NM_033453', mart = ensembl)

refseq_mrna chromosome_name transcript_start
1 NM_033453 20 3190006
transcript_end strand
1 3204516 1

library("biomaRt")
ensembl <- useMart("ensembl",dataset="hsapiens_gene_ensembl")
#ENST00000398344 is on chr22:24,313,554-24,316,773
getBM(attributes = c('chromosome_name',
                      'start_position',
                      'end_position',
                      'strand'
                     ),
                     filters = 'ensembl_transcript_id',
                     values = 'ENST00000398344',
                     mart = ensembl)
   chromosome_name start_position end_position strand
1              22       24313554     24322660     -1

# Make xref db

library(og.s.eg.db)
select( org.Hs.eg.db, keys=symbols, keytype="SYMBOL", column=c("SYMBOL", "REFSEQ") )
select( org.Hs.eg.db, keys=refseqs, keytype="REFSEQ", column=c("REFSEQ", "SYMBOL") )

library(biomaRt)
ensembl = useMart( "ensembl", dataset="hsapiens_gene_ensembl")

###-----------------------------------------------------------------------------
### gene start
###-----------------------------------------------------------------------------
geneCoord = getBM(attributes=c('hgnc_symbol', 'external_gene_id', 'chromosome_name','start_position','end_position','strand'), filters='hgnc_symbol', values=symbols, mart = ensembl)
geneCoord = geneCoord[ grep( "^[0-9]|^X|^Y", geneCoord$chromosome_name ), ]

###-----------------------------------------------------------------------------
### transcript_start
###----------------------------------------------------------------------------
txCoord = getBM(attributes=c('refseq_mrna', 'hgnc_symbol', 'chromosome_name','transcript_start','transcript_end','strand'), filters='refseq_mrna', values=refseqs, mart = ensembl)
txCoord = txCoord[ grep( "^[0-9]|^X|^Y", txCoord$chromosome_name ), ]


 

이메일로 전송BlogThis!X에 공유Facebook에서 공유

라벨: Informatics