Linear LP Gradient Analysis

Citation

If you use this code, please cite:

Baker MS, Heihre JM, Kulinski T, et al. The microdialysis-derived lactate–pyruvate gradient indicates anaerobic activity after brain injury. Manuscript under review.

---

Requirements

• Python 3.9+ (python.org)
• R 4.0+ (r-project.org)

Verify installation: python3 --version and R --version

---

Download

Navigate to where you want the project, then:

git clone https://github.com/msb-jr/linear-LP-gradient.git
cd linear-LP-gradient

Or click Code → Download ZIP on GitHub and extract.

---

Setup

Python Environment

python3 -m venv venv              # Create (first time only)
source venv/bin/activate          # Activate (Mac/Linux)
# venv\Scripts\activate.bat       # Activate (Windows)
pip install -r requirements.txt   # Install packages

R Environment

Rscript -e "renv::restore()"

---

Pipeline

Input Data

Place your data file at:

0_input/original_data.csv

Expected content: CSV with the following columns^a:

Column	Type	Unit	Description
patient	string or numeric	—	Unique patient identifier
time_since_injury	numeric	hours	Time elapsed since injury
lactate	numeric	mM	Cerebral microdialysis lactate
pyruvate	numeric	μM	Cerebral microdialysis pyruvate
glucoseb	numeric	mM	Cerebral microdialysis glucose
gosec	integer (1-8)	—	Glasgow Outcome Scale-Extended

Note: lpr (lactate/pyruvate ratio) is computed automatically by Script 1 from lactate and pyruvate.
^a Columns can contain missing values, they will be handled automatically as needed.
^b Used in Script 8
^c Used in Script 9

Additionally, Script 10 requires a separate in vitro dataset at 0_input/in_vitro.csv with columns: type (wt|rot), time (hours), lactate (mM), pyruvate (mM).

---

Scripts

Run scripts 1 → 2 in order, then can run 3+. See each script header for full details.

Script	Description	Manuscript relevance
1_pre-processing.py	Data pre-processing	2.2, SM2.ii
2_linear_segmentation.py	PELT linear segmentation + annotation	2.3.ii
3_segmentation_visual.py	Per-patient segmentation plots	Fig. 2, Fig. S2, Fig. S3
4_segmentation_stats.R	Segmentation yield summary	3.1, Table S1, Table S4
5_linear_LP_models.R	Linear model summaries	3.2, 3.3.i, Fig. 3
6_hyperbolic_LPR_error.R	LPR error analysis	3.3.ii, Fig. 4, Table S2, Table S5
7_never_always_LPR25.R	LPR cohort comparisons	3.4, Table S3
8_substrate_lpr_fidelity.R	Substrate delivery and LPR fidelity	3.5, Fig. 5
9_outcome.R	Outcome associations (GOSE)	3.6, Fig. 6
10_in_vitro.R	In vitro ETC inhibition validation	3.8, Fig. 7
extra_linear_to_hyperbolic.py	Conceptual linear→hyperbolic relationship	2.3.i, Fig. S1
extra_lpr_m_schematic.py	CMD Schematic with LPR timecourse & linear regressions	Fig. 1

Run

# Individual
python 1_pre-processing.py
python 2_linear_segmentation.py
# ... etc

# All at once
python 1_pre-processing.py && \
python 2_linear_segmentation.py && \
python 3_segmentation_visual.py && \
Rscript 4_segmentation_stats.R && \
Rscript 5_linear_LP_models.R && \
Rscript 6_hyperbolic_LPR_error.R && \
Rscript 7_never_always_LPR25.R && \
Rscript 8_substrate_lpr_fidelity.R && \
Rscript 9_outcome.R && \
Rscript 10_in_vitro.R && \
python extra_linear_to_hyperbolic.py && \
python extra_lpr_m_schematic.py

Outputs are saved to 1_output/<script_name>/.

---

Cleaning Up

Remove project environments:

• Python: delete the venv/ folder
• R: delete the renv/library/ folder