## Bioinformatik für Tier- und Pflanzenwissenschaften 
## VL-12 Databases
## Antworten zur Uebung
## 

# 1) UniprotKB
# a)
# https://www.uniprot.org/help/query-fields
# b)
# (organism_id:9031) AND (reviewed:true) # 2308
# c)
# (organism_id:9031) AND (reviewed:true) AND (date_created:[2011-09-01 TO *])  # 98
# d)
# O18836  GDF8_BOVIN
# e)
# Name:MSTN
# Synonyms:GDF8, MH
# f)
# ENSBTAT00000015674, ENSBTAP00000015674, ENSBTAG00000011808
# g)
# https://rest.uniprot.org/uniprotkb/O18836.txt
# DT: date, OS: organism Species, DR: Database cross-reference, FT: feature
# h) 
# Bos gaurus, Bos indicus, Bubalus bubalis (Domestic water buffalo), Taurotragus derbianus (Giant eland) (Derby eland),
# Pig, Pronghorn


# 2) biomaRt
# a)
if (!require("BiocManager", quietly = TRUE))
	install.packages("BiocManager")
BiocManager::install("biomaRt")
library(biomaRt)

# b)
mart <- useMart("ensembl")

# c)
listDatasets(mart)   
dim(listDatasets(mart)) # 215 datasets

# d)
mart <- useMart("ensembl", "mmusculus_gene_ensembl")

# e)
listAttributes(mart)  
dim(listAttributes(mart)) # 2972

# f)
# nur getBM() angucken

# g)
getBM(attributes = c("chromosome_name"), filters = "mgi_symbol", values = "Mstn", mart = mart)
getBM(attributes = c("chromosome_name"), values = "Mstn", mart = mart)

# h)
getBM(attributes = c("chromosome_name", "start_position", "end_position"), filters = "mgi_symbol", 
      values = "Mstn", mart = mart)

#   mgi_symbol chromosome_name start_position end_position
# 1       Mstn               1       53100799     53107238

# i)
# 53100799 - 100000 = 53000799
# 53107238 + 100000 = 53207238

getBM(attributes = c("mgi_symbol", "chromosome_name", "start_position", "end_position"), 
	  filters = c("chromosomal_region"), values = "1:53000799:53207238", mart = mart)

# 3 genes without Mstn

# mgi_symbol chromosome_name start_position end_position
# 1 1700019D03Rik               1       52961483     53059338
# 2          Mstn               1       53100799     53107238
# 3       Gm24349               1       53139209     53139309
# 4 1700019A02Rik               1       53197736     53226795