wehub-resource-sync
149 lines
3.4 KiB
YAML
149 lines
3.4 KiB
YAML
# Organism to Genome Mapping for nf-core Pipelines
|
|
# Maps organism names (as they appear in GEO/SRA) to iGenomes keys
|
|
|
|
organisms:
|
|
# Human
|
|
"Homo sapiens":
|
|
genome: "GRCh38"
|
|
taxid: 9606
|
|
aliases: ["human", "hg38", "GRCh38"]
|
|
notes: "Primary human reference genome"
|
|
|
|
"Homo sapiens (legacy)":
|
|
genome: "GRCh37"
|
|
taxid: 9606
|
|
aliases: ["hg19", "GRCh37"]
|
|
notes: "Legacy human reference, still used for some clinical data"
|
|
|
|
# Mouse
|
|
"Mus musculus":
|
|
genome: "GRCm39"
|
|
taxid: 10090
|
|
aliases: ["mouse", "mm39", "GRCm39"]
|
|
notes: "Current mouse reference genome"
|
|
|
|
"Mus musculus (legacy)":
|
|
genome: "GRCm38"
|
|
taxid: 10090
|
|
aliases: ["mm10", "GRCm38"]
|
|
notes: "Legacy mouse reference"
|
|
|
|
# Yeast
|
|
"Saccharomyces cerevisiae":
|
|
genome: "R64-1-1"
|
|
taxid: 4932
|
|
aliases: ["yeast", "sacCer3", "S288C", "budding yeast"]
|
|
notes: "S288C reference strain"
|
|
|
|
# Fruit fly
|
|
"Drosophila melanogaster":
|
|
genome: "BDGP6"
|
|
taxid: 7227
|
|
aliases: ["fly", "dm6", "fruit fly", "Dmel"]
|
|
notes: "Berkeley Drosophila Genome Project release 6"
|
|
|
|
# Worm
|
|
"Caenorhabditis elegans":
|
|
genome: "WBcel235"
|
|
taxid: 6239
|
|
aliases: ["worm", "ce11", "C. elegans", "Cele"]
|
|
notes: "WormBase reference"
|
|
|
|
# Zebrafish
|
|
"Danio rerio":
|
|
genome: "GRCz11"
|
|
taxid: 7955
|
|
aliases: ["zebrafish", "danRer11", "Drer"]
|
|
notes: "Genome Reference Consortium Zebrafish Build 11"
|
|
|
|
# Arabidopsis
|
|
"Arabidopsis thaliana":
|
|
genome: "TAIR10"
|
|
taxid: 3702
|
|
aliases: ["arabidopsis", "thale cress", "Atha"]
|
|
notes: "The Arabidopsis Information Resource v10"
|
|
|
|
# Rat
|
|
"Rattus norvegicus":
|
|
genome: "Rnor_6.0"
|
|
taxid: 10116
|
|
aliases: ["rat", "rn6", "Rnor"]
|
|
notes: "Rnor 6.0 reference"
|
|
|
|
# Chicken
|
|
"Gallus gallus":
|
|
genome: "GRCg6a"
|
|
taxid: 9031
|
|
aliases: ["chicken", "galGal6", "Ggal"]
|
|
notes: "Genome Reference Consortium Chicken Build 6a"
|
|
|
|
# Pig
|
|
"Sus scrofa":
|
|
genome: "Sscrofa11.1"
|
|
taxid: 9823
|
|
aliases: ["pig", "susScr11", "Sscr"]
|
|
notes: "Swine genome assembly 11.1"
|
|
|
|
# Cow
|
|
"Bos taurus":
|
|
genome: "ARS-UCD1.2"
|
|
taxid: 9913
|
|
aliases: ["cow", "bosTau9", "cattle", "Btau"]
|
|
notes: "USDA ARS assembly"
|
|
|
|
# Dog
|
|
"Canis lupus familiaris":
|
|
genome: "CanFam3.1"
|
|
taxid: 9615
|
|
aliases: ["dog", "canFam3", "Clup"]
|
|
notes: "Broad Institute CanFam3.1"
|
|
|
|
# Frog
|
|
"Xenopus tropicalis":
|
|
genome: "JGI_4.2"
|
|
taxid: 8364
|
|
aliases: ["frog", "xenTro9", "Xtro"]
|
|
notes: "JGI assembly version 4.2"
|
|
|
|
# Maize/Corn
|
|
"Zea mays":
|
|
genome: "Zm-B73-REFERENCE-NAM-5.0"
|
|
taxid: 4577
|
|
aliases: ["maize", "corn", "Zmay"]
|
|
notes: "B73 reference genome v5"
|
|
|
|
# Rice
|
|
"Oryza sativa":
|
|
genome: "IRGSP-1.0"
|
|
taxid: 39947
|
|
aliases: ["rice", "Osat"]
|
|
notes: "International Rice Genome Sequencing Project"
|
|
|
|
# E. coli (common bacterial model)
|
|
"Escherichia coli":
|
|
genome: null
|
|
taxid: 562
|
|
aliases: ["E. coli", "Ecol"]
|
|
notes: "Use specific strain reference; K-12 MG1655 common"
|
|
|
|
# Fission yeast
|
|
"Schizosaccharomyces pombe":
|
|
genome: "ASM294v2"
|
|
taxid: 4896
|
|
aliases: ["fission yeast", "S. pombe", "Spom"]
|
|
notes: "PomBase reference"
|
|
|
|
# Pipeline mapping based on library strategy
|
|
pipeline_suggestions:
|
|
"RNA-SEQ": "rnaseq"
|
|
"ATAC-SEQ": "atacseq"
|
|
"CHIP-SEQ": "chipseq"
|
|
"WGS": "sarek"
|
|
"WXS": "sarek"
|
|
"EXOME": "sarek"
|
|
"AMPLICON": "ampliseq"
|
|
"BISULFITE-SEQ": "methylseq"
|
|
"HI-C": "hic"
|
|
"MIRNA-SEQ": "smrnaseq"
|
|
"RRBS": "methylseq"
|