monsoon-nlp commited on
Commit
ad1e504
1 Parent(s): 2125e20

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +74 -0
README.md CHANGED
@@ -1,3 +1,77 @@
1
  ---
2
  license: cc
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc
3
+ tags:
4
+ - protein
5
+ - biology
6
+ - botany
7
+ pretty_name: GreenBeing
8
  ---
9
+
10
+ # GreenBeing Proteins dataset
11
+
12
+ Proteins from UniProtKB (knowledge base), which are from food crops and related species.
13
+
14
+ Amino acid sequences use [IUPAC-IUB codes](https://en.wikipedia.org/wiki/FASTA_format#Sequence_representation) where letters A-Z map to amino acids.
15
+
16
+ XML source from https://www.uniprot.org/help/downloads
17
+
18
+ CoLab notebook: https://colab.research.google.com/drive/1M6sO0Ws6i5z9VUXIXopiOqo1OkQ7K-1g?usp=sharing
19
+
20
+ ## Pretraining set (TODO)
21
+
22
+ Amino acid sequences for **unreviewed proteins (TrEMBL)**
23
+
24
+ Each row contains a species or subspecies name (for filtering), and the amino acid sequence.
25
+
26
+ ### Share of taxa
27
+
28
+ In the pretraining set:
29
+ - 31% Papilionoideae/Faboideae (soybeans, peas, pulses, peanuts)
30
+ - 19% Triticeae (wheat, barley, rye, and relatives)
31
+ - 17% Oryzeae (rice, wild rice, and relatives)
32
+ - 12% Paniceae (most millets)
33
+ - 6% Solanum (potatoes, tomatoes, other nightshades, etc.)
34
+ - 5% Zea (corn/maize)
35
+ - 2% Sorghum
36
+ - 2% Lactuca sativa (lettuce)
37
+ - 2% Capsicum (chili peppers)
38
+ - 2% Cucurbita (squash, pumpkins)
39
+ - 0.8% Spinacia
40
+ - 0.7% Asparagus
41
+ - 0.2% Beta vulgaris (beet)
42
+ - 0.1% Bambusa (includes edible bamboo shoots)
43
+
44
+ ## Finetuning set
45
+
46
+ **Reviewed proteins (Swiss-Prot)** from above taxa. Each row contains a gene name, species or subspecies, an amino acid sequence, and comments / annotations available in UniProt.
47
+
48
+ A gene name may match multiple entries on UniProt from different accessions.
49
+
50
+ Annotations may be empty, or may include information such as:
51
+
52
+ - likely function, written in English
53
+ - location inside of cell (e.g. "Subcellular locations: Vacuole")
54
+ - locations in the plant (e.g. "Expressed in roots, stems")
55
+
56
+ Removed PubMed reference numbers to avoid training models to hallucinate PubMed references.
57
+
58
+ In the current state of plant genomics research, about half of the reviewed set are from rice and related species,
59
+ a fifth are from Papilionoideae/Faboideae, and seven taxa have less than 1%.
60
+
61
+ ## Research set (TODO)
62
+
63
+ Formatted like the finetuning set, but with proteins from quinoa and amaranth.
64
+
65
+
66
+
67
+ ## Limitations and Safety Notes
68
+
69
+ Proteins and review status on UniProt are from late March 2024.
70
+
71
+ You should at least pretrain on more than *Oryza sativa*, because its reviewed proteins are probably some of the most common/important.
72
+
73
+ The reviewed sequences are often similar to unreviewed sequences from related species and accessions, and are not suitable to 'predict' protein completion.
74
+
75
+ Species include inedible wild relatives.<br/>
76
+ Some people have allergic reactions to wheat/gluten, nightshades, maize, and other crops.<br/>
77
+ Chili peppers can be painfully spicy.