Coverage for src/instawell/figures/fig_01_raw.py: 0%

1import logging 

2 

3import pandas as pd 

4import plotly.express as px 

5 

6from instawell.core.exp_context import ExperimentContext 

7from instawell.core.steps import StepFiles 

8 

9# set logging level to INFO 

10 

11logger = logging.getLogger(__name__) 

12 

13 

14def create_figures_generator(ctx: ExperimentContext): 

15 """ 

16 Generates Plotly figures from grouped data. 

17 

18 This function groups the input DataFrame by 'ligand', 'protein', and 

19 'buffer', then yields a line plot figure for each group. 

20 

21 Args: 

22 ctx: An ExperimentContext object containing experiment details. 

23 

24 Yields: 

25 A Plotly figure object for each group. 

26 """ 

27 # Load the organized data 

28 organized_data_path = ctx.experiment_dir / StepFiles.INGESTED_DATA 

29 if not organized_data_path.exists(): 

30 raise FileNotFoundError(f"Organized data file not found: {organized_data_path}") 

31 data_df = pd.read_csv(organized_data_path) 

32 

33 # ensure the necessary columns are present 

34 required_columns = [ 

35 "Temperature", 

36 "well", 

37 "value", 

38 "ligand", 

39 "protein", 

40 "buffer", 

41 "concentration", 

42 "well_unqcond", 

43 ] 

44 for col in required_columns: 

45 if col not in data_df.columns: 

46 raise ValueError(f"Missing required column: {col} in the data.") 

47 grouped_data = data_df.groupby(["ligand", "protein", "buffer"]) 

48 

49 for (ligand, protein, buffer), group in grouped_data: 

50 fig = px.line( 

51 group, 

52 x="Temperature", 

53 y="value", 

54 color="well_unqcond", 

55 title=f"Raw Data for {ligand} and {protein} in {buffer} Buffer", 

56 ) 

57 yield fig