Landscape projection

This does surface interpolation to generate a VR landscape

# This ensures Plotly output works in multiple places:
# plotly_mimetype: VS Code notebook UI
# notebook: "Jupyter: Export to HTML" command in VS Code
# See https://plotly.com/python/renderers/#multiple-renderers
pio.renderers.default = "plotly_mimetype+notebook"

base_dir = os.path.dirname(sys.path[0])
data_dir = f"{base_dir}/data/output"

with open(f"{data_dir}/param_dict.json", "r") as param_json:
    param_dict = json.load(param_json)

param_dict

{'input_csv': 'md_QinCardoso23.csv',
 'dim_names': ['phate1', 'phate2'],
 'potency': 'CCAT',
 'dynamic': 'inverse_vvl',
 'cluster': 'curatedCLUST',
 'cluster_colours': {'TA 1': '#FF99CC',
  'Goblet / DCS': '#CC9900',
  'Late Enterocyte': '#006600',
  'proCSC': '#000099',
  'Early Enterocyte': '#00FF99',
  'revCSC': '#CC0000',
  'ER Stress': '#CC00FF',
  'CSC': '#6666FF',
  'TA 2': '#FF6666'},
 'condition': 'condition',
 'scores_csv': 'vr_QinCardoso23.csv'}

score_data = pd.read_csv(f"{data_dir}/{param_dict['scores_csv']}")

score_data

	phate1	phate2	CCAT	velocity_length	A	K	P	WENR	mac	fib	...	OWNsig_stemS_UCell	condition	inverse_vvl	dist	deg	dist_scaled	inv_deg_scaled	potency_med	dynamic_med	VR
0	-0.003110	-0.003080	0.339365	186.242097	0	0	0	0	0	0	...	0.221648	WT	0.147660	0.006520	8	0.095761	0.312500	0.418507	0.062690	0.377256
1	0.024983	0.029213	0.217284	152.058837	0	0	0	0	0	0	...	0.122389	WT	0.231605	0.017569	6	0.059681	0.666667	0.249275	0.113152	0.225022
2	0.037142	0.007396	0.102314	164.789267	0	0	0	0	0	0	...	0.155778	WT	0.196273	0.019348	6	0.226228	0.714286	0.220369	0.189912	0.202629
3	-0.021990	0.000884	0.592517	221.681973	0	0	0	0	0	0	...	0.085037	WT	0.087964	0.008693	5	0.095699	1.000000	0.680724	0.069009	0.613312
4	0.041909	0.035762	0.207627	208.885837	0	0	0	0	0	0	...	0.058852	WT	0.107182	0.047344	5	0.920014	1.000000	0.220369	0.189912	0.215805
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
5896	-0.003300	-0.006133	0.598417	197.806256	0	0	0	1	0	0	...	0.337778	WENR	0.125830	0.005085	5	0.009114	1.000000	0.690199	0.098587	0.621269
5897	-0.028562	0.010894	0.842009	145.233683	0	0	0	1	0	0	...	0.256611	WENR	0.253098	0.008261	5	0.080657	1.000000	0.779294	0.210717	0.703064
5898	-0.002911	-0.009425	0.689346	188.500721	0	0	0	1	0	0	...	0.283296	WENR	0.143186	0.007716	5	0.129462	1.000000	0.724684	0.164599	0.654346
5899	-0.024535	0.002836	0.838003	114.664223	0	0	0	1	0	0	...	0.203278	WENR	0.380759	0.007152	6	0.042140	0.694444	0.779294	0.210717	0.702252
5900	-0.013252	0.004670	0.815766	191.966684	0	0	0	1	0	0	...	0.388148	WENR	0.136525	0.015421	7	0.749223	0.510204	0.696763	0.105690	0.635005

5901 rows × 22 columns

---

source

generate_grid

 generate_grid (dataframe, param_dict, auto=True, nticks=48, d1min=None,
                d1max=None, d1tick=None, d2min=None, d2max=None,
                d2tick=None)

param_dict = generate_grid(score_data, param_dict, auto=True)

Automated grid computation
Added grids to parameter dictionary

Eventually, setting condition to None in param_dict will signify that vr scores are computed on the whole dataset.

For now though, we assume that they are always computed on a per sample/condition basis, and thus several landscapes need to be generated.

---

source

interpolate_landscape

 interpolate_landscape (dataframe, param_dict, trim=True)

param_dict

{'input_csv': 'md_QinCardoso23.csv',
 'dim_names': ['phate1', 'phate2'],
 'potency': 'CCAT',
 'dynamic': 'inverse_vvl',
 'cluster': 'curatedCLUST',
 'cluster_colours': {'TA 1': '#FF99CC',
  'Goblet / DCS': '#CC9900',
  'Late Enterocyte': '#006600',
  'proCSC': '#000099',
  'Early Enterocyte': '#00FF99',
  'revCSC': '#CC0000',
  'ER Stress': '#CC00FF',
  'CSC': '#6666FF',
  'TA 2': '#FF6666'},
 'condition': 'condition',
 'scores_csv': 'vr_QinCardoso23.csv',
 'grids': array([[[-0.03413235, -0.03413235, -0.03413235, ..., -0.03413235,
          -0.03413235, -0.03413235],
         [-0.03262914, -0.03262914, -0.03262914, ..., -0.03262914,
          -0.03262914, -0.03262914],
         [-0.03112593, -0.03112593, -0.03112593, ..., -0.03112593,
          -0.03112593, -0.03112593],
         ...,
         [ 0.04253158,  0.04253158,  0.04253158, ...,  0.04253158,
           0.04253158,  0.04253158],
         [ 0.04403479,  0.04403479,  0.04403479, ...,  0.04403479,
           0.04403479,  0.04403479],
         [ 0.045538  ,  0.045538  ,  0.045538  , ...,  0.045538  ,
           0.045538  ,  0.045538  ]],
 
        [[-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659],
         [-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659],
         [-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659],
         ...,
         [-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659],
         [-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659],
         [-0.0224318 , -0.0212241 , -0.02001639, ...,  0.03916118,
           0.04036889,  0.04157659]]])}

# Colour and axes scales should be set globally here
vr_max = round(score_data["VR"].max(),2)+0.02
vr_min = round(score_data["VR"].min(),2)-0.02

for i in score_data[param_dict["condition"]].unique():
    print(i)
    land_data = score_data.loc[score_data[param_dict["condition"]]==i]
    print(land_data.shape)

    vr_land = interpolate_landscape(land_data, param_dict)

    fig = go.Figure(
            data=[go.Surface(
                x=param_dict["grids"][0],y=param_dict["grids"][1],z=vr_land
            )])
    fig.update_traces(colorscale=[[0,"blue"],[0.25,"green"],[0.5,"yellow"],[0.75,"#5f3316"],[0.85,"#513b1b"],[1,"white"]],cmax=vr_max, cmin=vr_min)
    fig.add_scatter3d(
        x=land_data[param_dict["dim_names"][0]],
        y=land_data[param_dict["dim_names"][1]], 
        z=((0.1*land_data[param_dict["potency"]]+0.9*land_data["VR"])+0.012),
        mode="markers",
        marker=dict(
            size=4, 
            color=land_data[param_dict["cluster"]].map(param_dict["cluster_colours"]),
            opacity=0.42,
            # line=dict(width=2,
            #     color="darkgrey")
            )
        )
    fig.update_layout(
        title=None,
        template = "simple_white",
        scene = dict(
            xaxis = dict(nticks=5,title=""),
            yaxis = dict(nticks=5, title = ""),
            zaxis = dict(nticks=5, range=[vr_min,vr_max], title="")
            ),
        autosize=True,
        width=1000, height=1000,
        margin=dict(l=0, r=0, b=0, t=0),
        scene_aspectmode="cube",
        scene_camera=dict(
            up=dict(x=0, y=0, z=1),
            center=dict(x=0, y=0, z=0),
            eye=dict(x=1.5, y=1, z=2)
        )
        )
    fig.show()

WT
(1511, 22)
CRC
(1860, 22)
WTfib
(1139, 22)
WENR
(1391, 22)

---

source

foo

 foo ()

# test_eq(say_hi("Jeremy"), "Hi Jeremy!")