Using a multivalent binding model to infer antibody Fc species from systems serology data

This model helps us learn relationships between antibody Fc structural features and immune receptor interaction, effector cell recruitment, and disease outcome. See our manuscript.

Installation

This project uses Rye for dependency management.

You can clone the repository and install the dependencies by running the following commands:

git clone https://github.com/meyer-lab/mechanismSerology.git
cd mechanismSerology
rye sync

Running the code

Figure generation

The figures can be generated using:

rye run make all

or for a specific figure:

rye run make output/figure_X.svg

Using the model

The model can be used without any fine-tuning on new systems serology datasets. The model uses numerical optimization to infer its outputs and this is handled by the optimize_loss function.

from maserol.core import optimize_loss

# load data ...

# run inference
opts = assemble_options(data)
x, ctx = optimize_loss(data, **opts, return_reshaped_params=True)
# x contains the inferred parameters, including the inferred antibody abundances (as "Rtot")

# if you want the inferences as a pandas DataFrame
Rtot = Rtot_to_df(x["Rtot"], data, rcps=list(opts["rcps"]))

Using our datasets

All of our datasets can be accessed through the maserol.datasets module.

from maserol.datasets import Zohar, Kaplonek

zohar = Zohar()

zohar_data = zohar.get_detection_signal()

zohar_meta = zohar.get_metadata()