KuangshiAi Claude Sonnet 4.5 commited on
Commit
b2aed13
·
1 Parent(s): c153ec2

Update molecular cases with LFS tracking for large files

Browse files

- Add .xtc, .gro, .jpg, .jpeg to LFS tracking
- Reorganize molecular visualization workflows

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Files changed (26) hide show
  1. .gitattributes +4 -1
  2. eval_cases/molecular_vis/README.md +0 -29
  3. eval_cases/molecular_vis/eval_analysis_tasks.yaml +158 -0
  4. eval_cases/molecular_vis/eval_analysis_workflows.yaml +109 -178
  5. eval_cases/napari/{eval_visualization_workflows.yaml → eval_visualization_tasks.yaml} +0 -0
  6. molecular_vis/{workflows/curved-membrane → curved-membrane}/GS/curved-membrane_gs.png +0 -0
  7. molecular_vis/{workflows/curved-membrane → curved-membrane}/GS/curved-membrane_gs.vmd +0 -0
  8. molecular_vis/curved-membrane/data/curved-membrane.gro +3 -0
  9. molecular_vis/{workflows/curved-membrane → curved-membrane}/task_description.txt +0 -0
  10. molecular_vis/{workflows/curved-membrane → curved-membrane}/visualization_goals.txt +0 -0
  11. molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/GS/ras-raf-membrane_gs.png +0 -0
  12. molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/GS/ras-raf-membrane_gs.vmd +0 -0
  13. molecular_vis/ras-raf-membrane/data/ras-raf-membrane.gro +3 -0
  14. molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/task_description.txt +0 -0
  15. molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/visualization_goals.txt +0 -0
  16. molecular_vis/trajectory-inspection/GS/avg_2d_dens_CDL0.png +3 -0
  17. molecular_vis/trajectory-inspection/GS/avg_2d_dens_POPC.png +3 -0
  18. molecular_vis/trajectory-inspection/GS/membrane-curved-tilted-side-5us.jpg +3 -0
  19. molecular_vis/trajectory-inspection/GS/relative_enrichment.png +3 -0
  20. molecular_vis/trajectory-inspection/GS/trajectory-inspection_gs.png +3 -0
  21. molecular_vis/trajectory-inspection/data/trajectory-inspection.gro +3 -0
  22. molecular_vis/trajectory-inspection/data/trajectory-inspection_3to5us.xtc +3 -0
  23. molecular_vis/trajectory-inspection/task_description.txt +41 -0
  24. molecular_vis/trajectory-inspection/visualization_goals.txt +3 -0
  25. molecular_vis/workflows/curved-membrane/data/curved-membrane.gro +0 -0
  26. molecular_vis/workflows/ras-raf-membrane/data/ras-raf-membrane.gro +0 -0
.gitattributes CHANGED
@@ -13,4 +13,7 @@
13
  *.cif filter=lfs diff=lfs merge=lfs -text
14
  *.mp4 filter=lfs diff=lfs merge=lfs -text
15
  *.nc filter=lfs diff=lfs merge=lfs -text
16
- bioimage_data/GS/dataset_002_zoom.jpg filter=lfs diff=lfs merge=lfs -text
 
 
 
 
13
  *.cif filter=lfs diff=lfs merge=lfs -text
14
  *.mp4 filter=lfs diff=lfs merge=lfs -text
15
  *.nc filter=lfs diff=lfs merge=lfs -text
16
+ *.xtc filter=lfs diff=lfs merge=lfs -text
17
+ *.gro filter=lfs diff=lfs merge=lfs -text
18
+ *.jpg filter=lfs diff=lfs merge=lfs -text
19
+ *.jpeg filter=lfs diff=lfs merge=lfs -text
eval_cases/molecular_vis/README.md DELETED
@@ -1,29 +0,0 @@
1
- # SciVisAgentBench: Molecular Visualization
2
-
3
- This benchmark is designed to evaluate agents on molecular visualization tasks, specifically focusing on tools like VMD (Visual Molecular Dynamics).
4
-
5
- ## Overview
6
-
7
- The benchmark assesses agent capabilities across different levels of complexity:
8
- - **Basic Actions (Easy)**: Simple, atomic commands and operations (Currently implemented).
9
- - **Workflows (Medium)**: Sequences of actions forming a coherent pipeline (Planned).
10
- - **Scientific Tasks (Hard)**: Complex, goal-oriented scientific analysis and visualization (Planned).
11
-
12
- ## Setup
13
-
14
- 1. **Data Preparation**:
15
- - Download the PDB file `1CRN` from the [RCSB PDB](https://www.rcsb.org/structure/1CRN).
16
- - Place the downloaded file into the `data/` folder in this directory.
17
-
18
- ## Usage
19
-
20
- You can run the evaluation cases defined in the YAML files (located in the `actions/` directory) using either:
21
- - **Promptfoo**
22
- - The **SciVisAgentBench Evaluation Framework**
23
-
24
- Refer to the [main repository](https://github.com/KuangshiAi/SciVisAgentBench/tree/main) for detailed instructions on running the evaluation harness.
25
-
26
- ## Future Extensions
27
-
28
- ### SciVisAgentBench: MD Simulations
29
- A placeholder for Molecular Dynamics (MD) simulation tasks is currently reserved. This will be expanded to include setting up, running, and analyzing simulations.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eval_cases/molecular_vis/eval_analysis_tasks.yaml ADDED
@@ -0,0 +1,158 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Test Cases for SciVisAgentBench molecular visualization workflows
2
+ # This test evaluates the ability to complete molecular visualization tasks
3
+ # with detailed requirements and evaluation criteria
4
+
5
+ #simple licorice visualization of a protein
6
+ - vars:
7
+ question: |
8
+ 1. I want you to visualize a molecular structure from a CIF file.
9
+ 2. Load the data/1CRN.cif.
10
+ 3. Visualize the molecular using a licorice representation.
11
+ 4. Take a screenshot of the visualization.
12
+ Q1. Does it show a licorice representation of the protein? (yes/no)
13
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_basic_vis.txt".
14
+ assert:
15
+ - type: llm-rubric
16
+ subtype: text
17
+ value: |
18
+ 1. Q1 correct answer: Yes
19
+ rs-file: "md_analysis/results/answers_basic_vis.txt"
20
+
21
+ #simple coloring by element of a protein
22
+ - vars:
23
+ question: |
24
+ 1. I want you to visualize a molecular structure from a CIF file.
25
+ 2. Load the data/1CRN.cif.
26
+ 3. Visualize the molecular using a CPK or similar representation where atoms are colored by their chemical element.
27
+ 4. Take a screenshot of the visualization.
28
+ Q1. Is the molecule colored according to the chemical element of its atoms (e.g., CPK coloring)? (yes/no)
29
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_element_coloring.txt".
30
+ assert:
31
+ - type: llm-rubric
32
+ subtype: text
33
+ value: |
34
+ 1. Q1 correct answer: Yes
35
+ rs-file: "md_analysis/results/answers_element_coloring.txt"
36
+
37
+ #simple selection and coloring of a protein
38
+ - vars:
39
+ question: |
40
+ 1. I want you to visualize a molecular structure from a CIF file.
41
+ 2. Load the data/1CRN.cif.
42
+ 3. Select all carbon atoms and color them cyan.
43
+ 4. Take a screenshot of the visualization.
44
+ Q1. Are all carbon atoms colored cyan? (yes/no)
45
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_selection_coloring.txt".
46
+ assert:
47
+ - type: llm-rubric
48
+ subtype: text
49
+ value: |
50
+ 1. Q1 correct answer: Yes
51
+ rs-file: "md_analysis/results/answers_selection_coloring.txt"
52
+
53
+ #simple coloring by charge of a protein
54
+ - vars:
55
+ question: |
56
+ 1. I want you to visualize a molecular structure from a CIF file.
57
+ 2. Load the data/1CRN.cif.
58
+ 3. Color the molecule according to atomic charge: use one color for positive charges, another for negative charges, and a third for neutral atoms.
59
+ 4. Take a screenshot of the visualization.
60
+ Q1. Is the molecule colored by atomic charge (differentiating positive, negative, and neutral)? (yes/no)
61
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_charge_coloring.txt".
62
+ assert:
63
+ - type: llm-rubric
64
+ subtype: text
65
+ value: |
66
+ 1. Q1 correct answer: Yes
67
+ rs-file: "md_analysis/results/answers_charge_coloring.txt"
68
+
69
+ #simple selection and coloring of specific atoms
70
+ - vars:
71
+ question: |
72
+ 1. I want you to visualize a molecular structure from a CIF file.
73
+ 2. Load the data/1CRN.cif.
74
+ 3. Select all oxygen atoms in residues 1 to 20 and color them red.
75
+ 4. Take a screenshot of the visualization.
76
+ Q1. Are all oxygen atoms in residues 1 to 20 colored red? (yes/no)
77
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_complex_selection.txt".
78
+ assert:
79
+ - type: llm-rubric
80
+ subtype: text
81
+ value: |
82
+ 1. Q1 correct answer: Yes
83
+ rs-file: "md_analysis/results/answers_complex_selection.txt"
84
+
85
+ #simple selection and coloring of aromatic residues
86
+ - vars:
87
+ question: |
88
+ 1. I want you to visualize a molecular structure from a CIF file.
89
+ 2. Load the data/1CRN.cif.
90
+ 3. Select all aromatic residues (PHE, TYR, TRP) and color them purple.
91
+ 4. Take a screenshot of the visualization.
92
+ Q1. Are all aromatic residues (PHE, TYR, TRP) colored purple? (yes/no)
93
+ 5. Answer Q1 in a plain text file "md_analysis/results/answers_aromatic_selection.txt".
94
+ assert:
95
+ - type: llm-rubric
96
+ subtype: text
97
+ value: |
98
+ 1. Q1 correct answer: Yes
99
+ rs-file: "md_analysis/results/answers_aromatic_selection.txt"
100
+
101
+ #simple RMSD and RMSF calculation of a protein
102
+ - vars:
103
+ question: |
104
+ 1. I want you to perform a structural analysis on a molecular structure from a CIF file.
105
+ 2. Load the data/1CRN.cif.
106
+ 3. Calculate the Root Mean Square Deviation (RMSD) of the structure against itself.
107
+ 4. Calculate the Root Mean Square Fluctuation (RMSF) for the structure.
108
+ 5. Save the computed RMSD and RMSF values as plain text to "md_analysis/results/answers_rmsd_rmsf.txt".
109
+ assert:
110
+ - type: llm-rubric
111
+ subtype: text
112
+ value: |
113
+ 1. Does the output report the calculated RMSD?
114
+ 2. Does the output report the calculated RMSF values or state that it requires a trajectory?
115
+ rs-file: "md_analysis/results/answers_rmsd_rmsf.txt"
116
+
117
+ #simple radius of gyration calculation of a protein
118
+ - vars:
119
+ question: |
120
+ 1. I want you to calculate the compactness of a protein from a CIF file.
121
+ 2. Load the data/1CRN.cif.
122
+ 3. Calculate the Radius of Gyration (Rg) of the protein structure.
123
+ 4. Save the calculated Radius of Gyration as plain text to "md_analysis/results/answers_rg.txt".
124
+ assert:
125
+ - type: llm-rubric
126
+ subtype: text
127
+ value: |
128
+ 1. Does the output report a numeric value for the calculated Radius of Gyration?
129
+ rs-file: "md_analysis/results/answers_rg.txt"
130
+
131
+ - vars:
132
+ question: |
133
+ 1. I want you to calculate specific geometric properties of a molecular structure from a CIF file.
134
+ 2. Load the data/1CRN.cif.
135
+ 3. Calculate the distance between the alpha carbons of residue 1 and residue 10.
136
+ 4. Calculate the backbone dihedral angles (phi and psi) for residue 5.
137
+ 5. Save the computed distance and angles as plain text to "md_analysis/results/answers_distances_angles.txt".
138
+ assert:
139
+ - type: llm-rubric
140
+ subtype: text
141
+ value: |
142
+ 1. Does the output report the calculated distance between the alpha carbons of residue 1 and 10?
143
+ 2. Does the output report the calculated phi and psi dihedral angles for residue 5?
144
+ rs-file: "md_analysis/results/answers_distances_angles.txt"
145
+
146
+ #simple contact calculation of a protein
147
+ - vars:
148
+ question: |
149
+ 1. I want you to calculate the number of contacts in a folded protein from a CIF file.
150
+ 2. Load the data/1CRN.cif.
151
+ 3. Calculate the number of contacts within an 8 Angstrom cutoff.
152
+ 4. Save the total count of contacts as plain text to "md_analysis/results/answers_native_contacts.txt".
153
+ assert:
154
+ - type: llm-rubric
155
+ subtype: text
156
+ value: |
157
+ 1. Does the output report a numeric count of contacts?
158
+ rs-file: "md_analysis/results/answers_native_contacts.txt"
eval_cases/molecular_vis/eval_analysis_workflows.yaml CHANGED
@@ -1,163 +1,48 @@
1
- # Test Cases for SciVisAgentBench molecular visualization workflows
2
- # This test evaluates the ability to complete molecular visualization tasks
3
- # with detailed requirements and evaluation criteria
4
-
5
- #simple licorice visualization of a protein
6
- - vars:
7
- question: |
8
- 1. I want you to visualize a molecular structure from a CIF file.
9
- 2. Load the data/1CRN.cif.
10
- 3. Visualize the molecular using a licorice representation.
11
- 4. Take a screenshot of the visualization.
12
- Q1. Does it show a licorice representation of the protein? (yes/no)
13
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_basic_vis.txt".
14
- assert:
15
- - type: llm-rubric
16
- subtype: text
17
- value: |
18
- 1. Q1 correct answer: Yes
19
- rs-file: "md_analysis/results/answers_basic_vis.txt"
20
-
21
- #simple coloring by element of a protein
22
  - vars:
23
  question: |
24
- 1. I want you to visualize a molecular structure from a CIF file.
25
- 2. Load the data/1CRN.cif.
26
- 3. Visualize the molecular using a CPK or similar representation where atoms are colored by their chemical element.
27
- 4. Take a screenshot of the visualization.
28
- Q1. Is the molecule colored according to the chemical element of its atoms (e.g., CPK coloring)? (yes/no)
29
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_element_coloring.txt".
30
- assert:
31
- - type: llm-rubric
32
- subtype: text
33
- value: |
34
- 1. Q1 correct answer: Yes
35
- rs-file: "md_analysis/results/answers_element_coloring.txt"
36
 
37
- #simple selection and coloring of a protein
38
- - vars:
39
- question: |
40
- 1. I want you to visualize a molecular structure from a CIF file.
41
- 2. Load the data/1CRN.cif.
42
- 3. Select all carbon atoms and color them cyan.
43
- 4. Take a screenshot of the visualization.
44
- Q1. Are all carbon atoms colored cyan? (yes/no)
45
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_selection_coloring.txt".
46
- assert:
47
- - type: llm-rubric
48
- subtype: text
49
- value: |
50
- 1. Q1 correct answer: Yes
51
- rs-file: "md_analysis/results/answers_selection_coloring.txt"
52
 
53
- #simple coloring by charge of a protein
54
- - vars:
55
- question: |
56
- 1. I want you to visualize a molecular structure from a CIF file.
57
- 2. Load the data/1CRN.cif.
58
- 3. Color the molecule according to atomic charge: use one color for positive charges, another for negative charges, and a third for neutral atoms.
59
- 4. Take a screenshot of the visualization.
60
- Q1. Is the molecule colored by atomic charge (differentiating positive, negative, and neutral)? (yes/no)
61
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_charge_coloring.txt".
62
- assert:
63
- - type: llm-rubric
64
- subtype: text
65
- value: |
66
- 1. Q1 correct answer: Yes
67
- rs-file: "md_analysis/results/answers_charge_coloring.txt"
68
 
69
- #simple selection and coloring of specific atoms
70
- - vars:
71
- question: |
72
- 1. I want you to visualize a molecular structure from a CIF file.
73
- 2. Load the data/1CRN.cif.
74
- 3. Select all oxygen atoms in residues 1 to 20 and color them red.
75
- 4. Take a screenshot of the visualization.
76
- Q1. Are all oxygen atoms in residues 1 to 20 colored red? (yes/no)
77
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_complex_selection.txt".
78
  assert:
79
  - type: llm-rubric
80
- subtype: text
81
  value: |
82
- 1. Q1 correct answer: Yes
83
- rs-file: "md_analysis/results/answers_complex_selection.txt"
84
-
85
- #simple selection and coloring of aromatic residues
86
- - vars:
87
- question: |
88
- 1. I want you to visualize a molecular structure from a CIF file.
89
- 2. Load the data/1CRN.cif.
90
- 3. Select all aromatic residues (PHE, TYR, TRP) and color them purple.
91
- 4. Take a screenshot of the visualization.
92
- Q1. Are all aromatic residues (PHE, TYR, TRP) colored purple? (yes/no)
93
- 5. Answer Q1 in a plain text file "md_analysis/results/answers_aromatic_selection.txt".
94
- assert:
95
  - type: llm-rubric
96
  subtype: text
97
  value: |
98
  1. Q1 correct answer: Yes
99
- rs-file: "md_analysis/results/answers_aromatic_selection.txt"
100
-
101
- #simple RMSD and RMSF calculation of a protein
102
- - vars:
103
- question: |
104
- 1. I want you to perform a structural analysis on a molecular structure from a CIF file.
105
- 2. Load the data/1CRN.cif.
106
- 3. Calculate the Root Mean Square Deviation (RMSD) of the structure against itself.
107
- 4. Calculate the Root Mean Square Fluctuation (RMSF) for the structure.
108
- 5. Save the computed RMSD and RMSF values as plain text to "md_analysis/results/answers_rmsd_rmsf.txt".
109
- assert:
110
- - type: llm-rubric
111
- subtype: text
112
- value: |
113
- 1. Does the output report the calculated RMSD?
114
- 2. Does the output report the calculated RMSF values or state that it requires a trajectory?
115
- rs-file: "md_analysis/results/answers_rmsd_rmsf.txt"
116
-
117
- #simple radius of gyration calculation of a protein
118
- - vars:
119
- question: |
120
- 1. I want you to calculate the compactness of a protein from a CIF file.
121
- 2. Load the data/1CRN.cif.
122
- 3. Calculate the Radius of Gyration (Rg) of the protein structure.
123
- 4. Save the calculated Radius of Gyration as plain text to "md_analysis/results/answers_rg.txt".
124
- assert:
125
- - type: llm-rubric
126
- subtype: text
127
- value: |
128
- 1. Does the output report a numeric value for the calculated Radius of Gyration?
129
- rs-file: "md_analysis/results/answers_rg.txt"
130
-
131
- - vars:
132
- question: |
133
- 1. I want you to calculate specific geometric properties of a molecular structure from a CIF file.
134
- 2. Load the data/1CRN.cif.
135
- 3. Calculate the distance between the alpha carbons of residue 1 and residue 10.
136
- 4. Calculate the backbone dihedral angles (phi and psi) for residue 5.
137
- 5. Save the computed distance and angles as plain text to "md_analysis/results/answers_distances_angles.txt".
138
- assert:
139
- - type: llm-rubric
140
- subtype: text
141
- value: |
142
- 1. Does the output report the calculated distance between the alpha carbons of residue 1 and 10?
143
- 2. Does the output report the calculated phi and psi dihedral angles for residue 5?
144
- rs-file: "md_analysis/results/answers_distances_angles.txt"
145
 
146
- #simple contact calculation of a protein
147
- - vars:
148
- question: |
149
- 1. I want you to calculate the number of contacts in a folded protein from a CIF file.
150
- 2. Load the data/1CRN.cif.
151
- 3. Calculate the number of contacts within an 8 Angstrom cutoff.
152
- 4. Save the total count of contacts as plain text to "md_analysis/results/answers_native_contacts.txt".
153
- assert:
154
- - type: llm-rubric
155
- subtype: text
156
- value: |
157
- 1. Does the output report a numeric count of contacts?
158
- rs-file: "md_analysis/results/answers_native_contacts.txt"
159
 
160
- # 1. Curved membrane simulation
161
  # As part of a protocol paper for building membrane using the coarse-grained Martini 3 force field, https://doi.org/10.1016/bs.mie.2024.03.010, a tutorial was developed: https://bbs.llnl.gov/data/building-membranes-data
162
  # Tutorial 4 is on curved membranes which is used here as an example of bilayer visualization.
163
  - vars:
@@ -186,47 +71,93 @@
186
 
187
  # Q1 reason: A few water molecules can be seen a little into the membrane.
188
 
189
-
190
- # 2. RAS-RAF on a plasma membrane
191
- # The RAS protein is lipid anchored and sits on the inner leaflet of the plasma membrane where it can bind RAF.
192
- # This is a snapshot of a RAS protein bound to the RBD and CRD domains of RAF on top of an eight-component plasma membrane setup with the coarse-grained Martini 2 force field.
193
- # This simulation frame is an example from the following study https://doi.org/10.1073/pnas.2113297119 and can be found online at: https://bbs.llnl.gov/data/ras-lipid-dependent-dynamics-data
194
  - vars:
195
  question: |
196
- 1. Please load the Martini coarse-grained simulation file from "ras-raf-membrane/data/ras-raf-membrane.gro" into VMD. The simulations has a membrane and a RAS-RAF protein complex.
197
-
198
- 2. Use VMD to show a zoomed in side view of the membrane and center on the protein with the protein below the membrane.
199
- For the bilayer only show the PO4 lipids beads and ROH cholesterol bead and color them gray.
200
- Also show the protein back bone beads coloring RAS (resid 2 to 187) red and RAF (resid 188 to 329) blue.
201
- Take a screenshot of the visualization.
202
 
203
- 3. Analyze the visualization and answer the following questions:
204
- Q1: Are there any cholesterol head groups in the bilayer center? (yes/no)
205
 
206
- Q2: How many lipids are there within 1.5 nm of the RAF protein?
207
- A. 0
208
- B. 0-3
209
- C. 3-5
210
- D. >5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
211
 
212
- 4. Save your work:
213
- Save the VMD state as "ras-raf-membrane/results/{agent_mode}/ras-raf-membrane.vmd".
214
- Save the screenshot of the visualization as "ras-raf-membrane/results/{agent_mode}/ras-raf-membrane.png".
215
- Save the answers to the analysis questions in plain text as "ras-raf-membrane/results/{agent_mode}/answers.txt".
216
  assert:
217
  - type: llm-rubric
218
  subtype: vision
 
 
219
  value: |
220
- 1. Camera Position: Is the image centered on a protein complex composed of clearly distinguishable half red RAS and half blue RAF components?
221
- 2. Membrane Placement: Is a lipid bilayer clearly visible above the protein?
222
- 3. Correct Leaflet Ordering: Is the inner leaflet of the bilayer closer to the protein and the outer leaflet positioned farther away?
 
223
  - type: llm-rubric
224
- subtype: text
 
 
225
  value: |
226
- 1. Q1 correct answer: Yes
227
- 2. Q2 correct answer: C. 3-5
 
228
 
229
- # Q1 reason: quite some gray dots from cholesterol ROH headgroup bead can be seen in the center of the bilayer, so the answer is YES.
230
- # Q2 reason: if you run the flowing VMD selection command (name PO4 ROH) and within 15 of (index 3457 to 35009 and resid 188 to 329)
231
- # You will find 4 lipids head group beads, so the answer is C. 3-5
232
- # Note here it’s important to only count each lipid once, e.g. either use a single reference bead per lipid or center of mass, also selecting the protein can be tricky as the lipids and water also if resid’s in the 188 to 329 range so either exclude other molecule or sub select only the protein residues
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 1. RAS-RAF on a plasma membrane
2
+ # The RAS protein is lipid anchored and sits on the inner leaflet of the plasma membrane where it can bind RAF.
3
+ # This is a snapshot of a RAS protein bound to the RBD and CRD domains of RAF on top of an eight-component plasma membrane setup with the coarse-grained Martini 2 force field.
4
+ # This simulation frame is an example from the following study https://doi.org/10.1073/pnas.2113297119 and can be found online at: https://bbs.llnl.gov/data/ras-lipid-dependent-dynamics-data
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
  - vars:
6
  question: |
7
+ 1. Please load the Martini coarse-grained simulation file from "ras-raf-membrane/data/ras-raf-membrane.gro" into VMD. The simulations has a membrane and a RAS-RAF protein complex.
8
+
9
+ 2. Use VMD to show a zoomed in side view of the membrane and center on the protein with the protein below the membrane.
10
+ For the bilayer only show the PO4 lipids beads and ROH cholesterol bead and color them gray.
11
+ Also show the protein back bone beads coloring RAS (resid 2 to 187) red and RAF (resid 188 to 329) blue.
12
+ Take a screenshot of the visualization.
 
 
 
 
 
 
13
 
14
+ 3. Analyze the visualization and answer the following questions:
15
+ Q1: Are there any cholesterol head groups in the bilayer center? (yes/no)
 
 
 
 
 
 
 
 
 
 
 
 
 
16
 
17
+ Q2: How many lipids are there within 1.5 nm of the RAF protein?
18
+ A. 0
19
+ B. 0-3
20
+ C. 3-5
21
+ D. >5
 
 
 
 
 
 
 
 
 
 
22
 
23
+ 4. Save your work:
24
+ Save the VMD state as "ras-raf-membrane/results/{agent_mode}/ras-raf-membrane.vmd".
25
+ Save the screenshot of the visualization as "ras-raf-membrane/results/{agent_mode}/ras-raf-membrane.png".
26
+ Save the answers to the analysis questions in plain text as "ras-raf-membrane/results/{agent_mode}/answers.txt".
 
 
 
 
 
27
  assert:
28
  - type: llm-rubric
29
+ subtype: vision
30
  value: |
31
+ 1. Camera Position: Is the image centered on a protein complex composed of clearly distinguishable half red RAS and half blue RAF components?
32
+ 2. Membrane Placement: Is a lipid bilayer clearly visible above the protein?
33
+ 3. Correct Leaflet Ordering: Is the inner leaflet of the bilayer closer to the protein and the outer leaflet positioned farther away?
 
 
 
 
 
 
 
 
 
 
34
  - type: llm-rubric
35
  subtype: text
36
  value: |
37
  1. Q1 correct answer: Yes
38
+ 2. Q2 correct answer: C. 3-5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
39
 
40
+ # Q1 reason: quite some gray dots from cholesterol ROH headgroup bead can be seen in the center of the bilayer, so the answer is YES.
41
+ # Q2 reason: if you run the flowing VMD selection command (name PO4 ROH) and within 15 of (index 3457 to 35009 and resid 188 to 329)
42
+ # You will find 4 lipids head group beads, so the answer is C. 3-5
43
+ # Note here it’s important to only count each lipid once, e.g. either use a single reference bead per lipid or center of mass, also selecting the protein can be tricky as the lipids and water also if resid’s in the 188 to 329 range so either exclude other molecule or sub select only the protein residues
 
 
 
 
 
 
 
 
 
44
 
45
+ # 2. Curved membrane simulation
46
  # As part of a protocol paper for building membrane using the coarse-grained Martini 3 force field, https://doi.org/10.1016/bs.mie.2024.03.010, a tutorial was developed: https://bbs.llnl.gov/data/building-membranes-data
47
  # Tutorial 4 is on curved membranes which is used here as an example of bilayer visualization.
48
  - vars:
 
71
 
72
  # Q1 reason: A few water molecules can be seen a little into the membrane.
73
 
74
+ # 3. Curved membrane trajectory inspection
 
 
 
 
75
  - vars:
76
  question: |
77
+ 1. Please load the Martini coarse-grained membrane simulation from
78
+ "trajectory-inspection/data/trajectory-inspection.gro" into VMD.
 
 
 
 
79
 
80
+ 2. Load the trajectory file
81
+ "trajectory-inspection/data/trajectory-inspection_3to5us.xtc".
82
 
83
+ 3. Render a titled side view of the membrane using the last frame of the trajectory.
84
+
85
+ - Show the PO4 beads of POPC lipids in light brown.
86
+ - Show the PO4 beads of CDL0 lipids in red.
87
+ - Display the simulation box in blue.
88
+
89
+ Save the rendered image as:
90
+ "trajectory-inspection/results/{agent_mode}/membrane-curved-tilted-side-5us.jpg"
91
+
92
+ 4. Perform curvature-based lipid distribution analysis and generate the following figures:
93
+
94
+ - A 2D density heatmap of CDL0 lipids across the curved membrane surface.
95
+ - A 2D density heatmap of POPC lipids across the curved membrane surface.
96
+ - A plot of lipid relative enrichment versus membrane mean curvature for both POPC and CDL0.
97
+
98
+ Save the generated figures as:
99
+
100
+ "trajectory-inspection/results/{agent_mode}/avg_2d_dens_CDL0.png"
101
+ "trajectory-inspection/results/{agent_mode}/avg_2d_dens_POPC.png"
102
+ "trajectory-inspection/results/{agent_mode}/relative_enrichment.png"
103
+
104
+ 5. Analyze the trajectory and answer the following questions:
105
+
106
+ Q1: Are there more than 3000 frames in the trajectory? (yes/no)
107
+
108
+ Q2: Is the ratio of POPC lipids to the neutral cardiolipin (CDL0) 8:1? (yes/no)
109
+
110
+ Q3: Do the CDL0 lipids enrich in the negatively curved membrane regions? (yes/no)
111
+
112
+ Q4: Does the total lipid density change significantly with membrane mean curvature? (yes/no)
113
+
114
+ Q5: Do the POPC lipids enrich in the negatively curved membrane regions? (yes/no)
115
+
116
+ 6. Save the answers to the analysis questions in plain text as
117
+ "trajectory-inspection/results/{agent_mode}/answers.txt".
118
 
 
 
 
 
119
  assert:
120
  - type: llm-rubric
121
  subtype: vision
122
+ gs_file: trajectory-inspection/GS/membrane-curved-tilted-side-5us.jpg
123
+ rs_file: trajectory-inspection/results/{agent_mode}/membrane-curved-tilted-side-5us.jpg
124
  value: |
125
+ 1. Do both images show a side view of a curved lipid bilayer rather than a top-down view?
126
+ 2. Are two lipid species visible with different colors, corresponding to POPC and CDL0 PO4 beads?
127
+ 3. Does the membrane visibly undulate, with one lipid type enriched on the inner curvature and the other on the outer curvature?
128
+
129
  - type: llm-rubric
130
+ subtype: vision
131
+ gs_file: trajectory-inspection/GS/avg_2d_dens_CDL0.png
132
+ rs_file: trajectory-inspection/results/{agent_mode}/avg_2d_dens_CDL0.png
133
  value: |
134
+ 1. Do both images show a curved membrane band visualized as a heatmap density map?
135
+ 2. Are sampling markers or dots visible along the membrane surface in both images?
136
+ 3. Is higher CDL0 density concentrated along the inner side of the curved membrane in both images?
137
 
138
+ - type: llm-rubric
139
+ subtype: vision
140
+ gs_file: trajectory-inspection/GS/avg_2d_dens_POPC.png
141
+ rs_file: trajectory-inspection/results/{agent_mode}/avg_2d_dens_POPC.png
142
+ value: |
143
+ 1. Do both images show a curved membrane band represented as a heatmap density map?
144
+ 2. Are sampling markers or dots visible along the membrane surface in both images?
145
+ 3. Is higher POPC density concentrated along the outer side of the curved membrane in both images?
146
+
147
+ - type: llm-rubric
148
+ subtype: vision
149
+ gs_file: trajectory-inspection/GS/relative_enrichment.png
150
+ rs_file: trajectory-inspection/results/{agent_mode}/relative_enrichment.png
151
+ value: |
152
+ 1. Do both plots show enrichment values plotted against membrane mean curvature?
153
+ 2. Are two curves corresponding to POPC and CDL0 visible in both plots?
154
+ 3. Is a horizontal reference line around enrichment = 1 present in both plots?
155
+
156
+ - type: llm-rubric
157
+ subtype: text
158
+ value: |
159
+ 1. Q1 correct answer: No (the trajectory contains 2000 frames).
160
+ 2. Q2 correct answer: No (the lipid ratio is 9:1).
161
+ 3. Q3 correct answer: Yes (CDL0 lipids enrich in negatively curved regions).
162
+ 4. Q4 correct answer: Yes (relative density varies roughly between 0.8 and 1.3).
163
+ 5. Q5 correct answer: No (POPC lipids enrich in positively curved regions, not negative).
eval_cases/napari/{eval_visualization_workflows.yaml → eval_visualization_tasks.yaml} RENAMED
File without changes
molecular_vis/{workflows/curved-membrane → curved-membrane}/GS/curved-membrane_gs.png RENAMED
File without changes
molecular_vis/{workflows/curved-membrane → curved-membrane}/GS/curved-membrane_gs.vmd RENAMED
File without changes
molecular_vis/curved-membrane/data/curved-membrane.gro ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13037008ccc807f7ad86da3e222a0d632841a3fbb1cd04b55cd0dcf3336ed67a
3
+ size 1106890
molecular_vis/{workflows/curved-membrane → curved-membrane}/task_description.txt RENAMED
File without changes
molecular_vis/{workflows/curved-membrane → curved-membrane}/visualization_goals.txt RENAMED
File without changes
molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/GS/ras-raf-membrane_gs.png RENAMED
File without changes
molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/GS/ras-raf-membrane_gs.vmd RENAMED
File without changes
molecular_vis/ras-raf-membrane/data/ras-raf-membrane.gro ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e7083794405852ca2a273ba9cd873cacb4f0327edbb373bb6bf222ff8f65ff3
3
+ size 9401161
molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/task_description.txt RENAMED
File without changes
molecular_vis/{workflows/ras-raf-membrane → ras-raf-membrane}/visualization_goals.txt RENAMED
File without changes
molecular_vis/trajectory-inspection/GS/avg_2d_dens_CDL0.png ADDED

Git LFS Details

  • SHA256: 4e3316739bf86f455eee3aec5d6d54e43463cad6b2d274d0af2cdace85521668
  • Pointer size: 131 Bytes
  • Size of remote file: 698 kB
molecular_vis/trajectory-inspection/GS/avg_2d_dens_POPC.png ADDED

Git LFS Details

  • SHA256: f44f263e0b3a23e763943584bb009664e9d6a43ab047f1e781c805b0fb53a0e4
  • Pointer size: 132 Bytes
  • Size of remote file: 1.04 MB
molecular_vis/trajectory-inspection/GS/membrane-curved-tilted-side-5us.jpg ADDED

Git LFS Details

  • SHA256: c92d349db053a8710e24c502ae1d0c278dde7c1dcd1654f26205b61dd46612e6
  • Pointer size: 131 Bytes
  • Size of remote file: 203 kB
molecular_vis/trajectory-inspection/GS/relative_enrichment.png ADDED

Git LFS Details

  • SHA256: 3e430c5c88f3c48eb30725739059f725645431c04a1e68a61cdf57bc4be5ea45
  • Pointer size: 131 Bytes
  • Size of remote file: 397 kB
molecular_vis/trajectory-inspection/GS/trajectory-inspection_gs.png ADDED

Git LFS Details

  • SHA256: c92d349db053a8710e24c502ae1d0c278dde7c1dcd1654f26205b61dd46612e6
  • Pointer size: 131 Bytes
  • Size of remote file: 203 kB
molecular_vis/trajectory-inspection/data/trajectory-inspection.gro ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:13037008ccc807f7ad86da3e222a0d632841a3fbb1cd04b55cd0dcf3336ed67a
3
+ size 1106890
molecular_vis/trajectory-inspection/data/trajectory-inspection_3to5us.xtc ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:219812a2447c7e03851cdcb83af4486a6a1dae069cd40d7912609f903fed277a
3
+ size 158624812
molecular_vis/trajectory-inspection/task_description.txt ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 1. Please load the Martini coarse-grained membrane simulation from
2
+ "trajectory-inspection/data/trajectory-inspection.gro" into VMD.
3
+
4
+ 2. Load the trajectory file
5
+ "trajectory-inspection/data/trajectory-inspection_3to5us.xtc".
6
+
7
+ 3. Render a titled side view of the membrane using the last frame of the trajectory.
8
+
9
+ - Show the PO4 beads of POPC lipids in light brown.
10
+ - Show the PO4 beads of CDL0 lipids in red.
11
+ - Display the simulation box in blue.
12
+
13
+ Save the rendered image as:
14
+ "trajectory-inspection/results/{agent_mode}/membrane-curved-tilted-side-5us.jpg"
15
+
16
+ 4. Perform curvature-based lipid distribution analysis and generate the following figures:
17
+
18
+ - A 2D density heatmap of CDL0 lipids across the curved membrane surface.
19
+ - A 2D density heatmap of POPC lipids across the curved membrane surface.
20
+ - A plot of lipid relative enrichment versus membrane mean curvature for both POPC and CDL0.
21
+
22
+ Save the generated figures as:
23
+
24
+ "trajectory-inspection/results/{agent_mode}/avg_2d_dens_CDL0.png"
25
+ "trajectory-inspection/results/{agent_mode}/avg_2d_dens_POPC.png"
26
+ "trajectory-inspection/results/{agent_mode}/relative_enrichment.png"
27
+
28
+ 5. Analyze the trajectory and answer the following questions:
29
+
30
+ Q1: Are there more than 3000 frames in the trajectory? (yes/no)
31
+
32
+ Q2: Is the ratio of POPC lipids to the neutral cardiolipin (CDL0) 8:1? (yes/no)
33
+
34
+ Q3: Do the CDL0 lipids enrich in the negatively curved membrane regions? (yes/no)
35
+
36
+ Q4: Does the total lipid density change significantly with membrane mean curvature? (yes/no)
37
+
38
+ Q5: Do the POPC lipids enrich in the negatively curved membrane regions? (yes/no)
39
+
40
+ 6. Save the answers to the analysis questions in plain text as
41
+ "trajectory-inspection/results/{agent_mode}/answers.txt".
molecular_vis/trajectory-inspection/visualization_goals.txt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ 1. Do both images show a side view of a curved lipid bilayer rather than a top-down view?
2
+ 2. Are two lipid species visible with different colors, corresponding to POPC and CDL0 PO4 beads?
3
+ 3. Does the membrane visibly undulate, with one lipid type enriched on the inner curvature and the other on the outer curvature?
molecular_vis/workflows/curved-membrane/data/curved-membrane.gro DELETED
The diff for this file is too large to render. See raw diff
 
molecular_vis/workflows/ras-raf-membrane/data/ras-raf-membrane.gro DELETED
The diff for this file is too large to render. See raw diff