jie
jie
is a spatial genome aligner. This package parses true chromatin imaging signal from noise by aligning signals to a reference DNA polymer model.
The codename is a tribute to the Chinese homophones:
结 (jié)
: a knot, a nod to the mysterious and often entangled structures of DNA解 (jiĕ)
: to solve, to untie, our bid to uncover these structures amid noise and uncertainty姐 (jiĕ)
: sister, our ability to resolve tightly paired replicated chromatids
Installation
Step 1 - Clone this repository:
git clone https://github.com/b2jia/jie.git cd jie
Step 2 - Create a new conda environment and install dependencies:
conda create --name jie -f environment.yml conda activate jie
Step 3 - Install jie
:
pip install -e .
To test, run:
python -W ignore test/test_jie.py
Usage
jie
is an exposition of chromatin tracing using polymer physics. The main function of this package is to illustrate the utility and power of spatial genome alignment.
jie
is NOT an all-purpose spatial genome aligner. Chromatin imaging is a nascent field and data collection is still being standardized. This aligner may not be compatible with different imaging protocols and data formats, among other variables.
We provide a vignette under jie/jupyter/
, with emphasis on inspectability
. This walks through the intuition of our spatial genome alignment and polymer fiber karyotyping routines:
00-spatial-genome-alignment-walk-thru.ipynb
We also provide a series of Jupyter notebooks (jie/jupyter/
), with emphasis on reproducibility
. This reproduces figures from our accompanying manuscript:
01-seqFISH-plus-mouse-ESC-spatial-genome-alignment.ipynb 02-seqFISH-plus-mouse-ESC-polymer-fiber-karyotyping.ipynb 03-seqFISH-plus-mouse-brain-spatial-genome-alignment.ipynb 04-seqFISH-plus-mouse-brain-polymer-fiber-karyotyping.ipynb 05-bench-mark-spatial-genome-agignment-against-chromatin-tracing-algorithm.ipynb
A command-line tool forthcoming.
Motivation
Multiplexed DNA-FISH is a powerful imaging technology that enables us to peer directly at the spatial location of genes inside the nucleus. Each gene appears as tiny dot under imaging.
Pivotally, figuring out which dots are physically linked would trace out the structure of chromosomes. Unfortunately, imaging is noisy, and single-cell biology is extremely variable. The two confound each other, making chromatin tracing prohibitively difficult!
For instance, in a diploid cell line with two copies of a gene we expect to see two spots. But what happens when we see:
-
-
Extra signals
: -
-
-
Is it
noise
? -
Off-target labeling
: The FISH probes might inadvertently label an off-target gene
-
Is it
-
-
Or is it
biological variation
? -
Aneuploidy
: A cell (ie. cancerous cell) may have more than one copy of a geneCell cycle
: When a cell gets ready to divide, it duplicates its genes
-
Or is it
-
-
-
-
Missing signals
: -
-
-
Is it
noise
? -
Poor probe labeling
: The FISH probes never labeled the intended target gene
-
Is it
-
-
Or is it
biological variation
? -
Copy Number Variation
: A cell may have a gene deletion
-
Or is it
-
-
If true signal and noise are indistinguishable, how do we know we are selecting true signals during chromatin tracing? It is not obvious which spots should be connected as part of a chromatin fiber. This dilemma was first aptly characterized by Ross et al. (https://journals.aps.org/pre/abstract/10.1103/PhysRevE.86.011918), which is nothing short of prescient...!
jie
is, conceptually, a spatial genome aligner that disambiguates spot selection by checking each imaged signal against a reference polymer physics model of chromatin. It relies on the key insight that the spatial separation
between two genes should be congruent
with its genomic separation
.
It makes no assumptions about the expected copy number of a gene, and when it traces chromatin it does so instead by evaluating the physical likelihood
of the chromatin fiber. In doing so, we can uncover copy number variations and even sister chromatids from multiplexed DNA-FISH imaging data.
Citation
Contact
Author: | Bojing (Blair) Jia |
---|---|
Email: | b2jia at eng dot ucsd dot edu |
Position: | MD-PhD Student, Ren Lab |
For other work related to single-cell biology, 3D genome, and chromatin imaging, please visit Prof. Bing Ren's website: http://renlab.sdsc.edu/