stefan-insilico
commited on
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,86 @@
|
|
1 |
-
---
|
2 |
-
license: cc-by-nc-4.0
|
3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cc-by-nc-4.0
|
3 |
+
---
|
4 |
+
|
5 |
+
# Precious3-GPT-Multi-Modal inference
|
6 |
+
|
7 |
+
Model inference is running at HuggingFace Inference endpoint
|
8 |
+
|
9 |
+
|
10 |
+
## Definitions
|
11 |
+
|
12 |
+
- **Signature**: up- and down-gene lists
|
13 |
+
|
14 |
+
---
|
15 |
+
|
16 |
+
## Generation config .json Overview
|
17 |
+
|
18 |
+
In the following example all possible configuration fields are specified. You can leave some meta-data fields in ```inputs``` section empty string(```""```) or empty list(```[]```).
|
19 |
+
|
20 |
+
For example, if you want to generate signature given specific meta-data you can use the following configuration. Note, ```up``` and ```down``` fields are empty lists as you want to generate them.
|
21 |
+
|
22 |
+
Another example - predict compound based on signature. You can take
|
23 |
+
|
24 |
+
```json
|
25 |
+
{
|
26 |
+
"inputs": {
|
27 |
+
"instruction": "compound2diff2compound",
|
28 |
+
"tissue": ["whole blood"],
|
29 |
+
"cell": "u937",
|
30 |
+
"efo": "Orphanet_139399",
|
31 |
+
"datatype": "",
|
32 |
+
"drug": "",
|
33 |
+
"dose": "",
|
34 |
+
"time": "",
|
35 |
+
"case": "",
|
36 |
+
"control": "",
|
37 |
+
"age": "",
|
38 |
+
"dataset_type": "expression",
|
39 |
+
"gender": "m",
|
40 |
+
"species": "human",
|
41 |
+
"up": [],
|
42 |
+
"down": []
|
43 |
+
},
|
44 |
+
"mode": "meta2diff",
|
45 |
+
"parameters": {
|
46 |
+
"temperature": 0.8,
|
47 |
+
"top_p": 0.2,
|
48 |
+
"top_k": 3550,
|
49 |
+
"n_next_tokens": 50
|
50 |
+
}
|
51 |
+
}
|
52 |
+
|
53 |
+
```
|
54 |
+
|
55 |
+
---
|
56 |
+
|
57 |
+
## Generation Modes (`mode` in config)
|
58 |
+
|
59 |
+
Choose the appropriate mode based on your requirements:
|
60 |
+
|
61 |
+
1. **meta2diff**: Generate signature given meta-data such as tissue, compound, gender, etc.
|
62 |
+
2. **diff2compound**: Predict compounds based on signature.
|
63 |
+
3. **meta2diff2compound**: Generate signatures given meta-data and then predict compounds based on generated signatures.
|
64 |
+
|
65 |
+
---
|
66 |
+
|
67 |
+
|
68 |
+
### Instruction (`inputs.instruction` in config)
|
69 |
+
|
70 |
+
You can use the following instructions (one or several at a time):
|
71 |
+
|
72 |
+
1. disease2diff2disease - generate signature for disease
|
73 |
+
2. compound2diff2compound - generate signature for compound
|
74 |
+
3. age_group2diff2age_group - generate signature for age-group
|
75 |
+
4. Under development:
|
76 |
+
- pathway2gene_list2pathway
|
77 |
+
- crispr_pert2diff2crispr_pert
|
78 |
+
- sh_crispr_pert2diff2sh_crispr_pert
|
79 |
+
- oe_lincs_pert2diff2oe_lincs_pert
|
80 |
+
|
81 |
+
|
82 |
+
### Other meta-data (`inputs.` in config)
|
83 |
+
Full list of available values for each meta-data item you can find in ```p3_entities_with_type.csv```
|
84 |
+
|
85 |
+
|
86 |
+
## Multi-Modality
|