Skills › Knowledge Work Plugins › nextflow-development › genomes.yaml
genomes.yaml
148 lines · bio-research/skills/nextflow-development/scripts/config/genomes.yaml at 2574e29
1# Organism to Genome Mapping for nf-core Pipelines
2# Maps organism names (as they appear in GEO/SRA) to iGenomes keys
4organisms:
5 # Human
6 "Homo sapiens":
7 genome: "GRCh38"
8 taxid: 9606
9 aliases: ["human", "hg38", "GRCh38"]
10 notes: "Primary human reference genome"
12 "Homo sapiens (legacy)":
13 genome: "GRCh37"
14 taxid: 9606
15 aliases: ["hg19", "GRCh37"]
16 notes: "Legacy human reference, still used for some clinical data"
18 # Mouse
19 "Mus musculus":
20 genome: "GRCm39"
21 taxid: 10090
22 aliases: ["mouse", "mm39", "GRCm39"]
23 notes: "Current mouse reference genome"
25 "Mus musculus (legacy)":
26 genome: "GRCm38"
27 taxid: 10090
28 aliases: ["mm10", "GRCm38"]
29 notes: "Legacy mouse reference"
31 # Yeast
32 "Saccharomyces cerevisiae":
33 genome: "R64-1-1"
34 taxid: 4932
35 aliases: ["yeast", "sacCer3", "S288C", "budding yeast"]
36 notes: "S288C reference strain"
38 # Fruit fly
39 "Drosophila melanogaster":
40 genome: "BDGP6"
41 taxid: 7227
42 aliases: ["fly", "dm6", "fruit fly", "Dmel"]
43 notes: "Berkeley Drosophila Genome Project release 6"
45 # Worm
46 "Caenorhabditis elegans":
47 genome: "WBcel235"
48 taxid: 6239
49 aliases: ["worm", "ce11", "C. elegans", "Cele"]
50 notes: "WormBase reference"
52 # Zebrafish
53 "Danio rerio":
54 genome: "GRCz11"
55 taxid: 7955
56 aliases: ["zebrafish", "danRer11", "Drer"]
57 notes: "Genome Reference Consortium Zebrafish Build 11"
59 # Arabidopsis
60 "Arabidopsis thaliana":
61 genome: "TAIR10"
62 taxid: 3702
63 aliases: ["arabidopsis", "thale cress", "Atha"]
64 notes: "The Arabidopsis Information Resource v10"
66 # Rat
67 "Rattus norvegicus":
68 genome: "Rnor_6.0"
69 taxid: 10116
70 aliases: ["rat", "rn6", "Rnor"]
71 notes: "Rnor 6.0 reference"
73 # Chicken
74 "Gallus gallus":
75 genome: "GRCg6a"
76 taxid: 9031
77 aliases: ["chicken", "galGal6", "Ggal"]
78 notes: "Genome Reference Consortium Chicken Build 6a"
80 # Pig
81 "Sus scrofa":
82 genome: "Sscrofa11.1"
83 taxid: 9823
84 aliases: ["pig", "susScr11", "Sscr"]
85 notes: "Swine genome assembly 11.1"
87 # Cow
88 "Bos taurus":
89 genome: "ARS-UCD1.2"
90 taxid: 9913
91 aliases: ["cow", "bosTau9", "cattle", "Btau"]
92 notes: "USDA ARS assembly"
94 # Dog
95 "Canis lupus familiaris":
96 genome: "CanFam3.1"
97 taxid: 9615
98 aliases: ["dog", "canFam3", "Clup"]
99 notes: "Broad Institute CanFam3.1"
101 # Frog
102 "Xenopus tropicalis":
103 genome: "JGI_4.2"
104 taxid: 8364
105 aliases: ["frog", "xenTro9", "Xtro"]
106 notes: "JGI assembly version 4.2"
108 # Maize/Corn
109 "Zea mays":
110 genome: "Zm-B73-REFERENCE-NAM-5.0"
111 taxid: 4577
112 aliases: ["maize", "corn", "Zmay"]
113 notes: "B73 reference genome v5"
115 # Rice
116 "Oryza sativa":
117 genome: "IRGSP-1.0"
118 taxid: 39947
119 aliases: ["rice", "Osat"]
120 notes: "International Rice Genome Sequencing Project"
122 # E. coli (common bacterial model)
123 "Escherichia coli":
124 genome: null
125 taxid: 562
126 aliases: ["E. coli", "Ecol"]
127 notes: "Use specific strain reference; K-12 MG1655 common"
129 # Fission yeast
130 "Schizosaccharomyces pombe":
131 genome: "ASM294v2"
132 taxid: 4896
133 aliases: ["fission yeast", "S. pombe", "Spom"]
134 notes: "PomBase reference"
136# Pipeline mapping based on library strategy
137pipeline_suggestions:
138 "RNA-SEQ": "rnaseq"
139 "ATAC-SEQ": "atacseq"
140 "CHIP-SEQ": "chipseq"
141 "WGS": "sarek"
142 "WXS": "sarek"
143 "EXOME": "sarek"
144 "AMPLICON": "ampliseq"
145 "BISULFITE-SEQ": "methylseq"
146 "HI-C": "hic"
147 "MIRNA-SEQ": "smrnaseq"
148 "RRBS": "methylseq"