# Unfolding the genome: the case study of P. falciparum
by Nelle Varoquaux
This git repository contains all the scripts to reproduce the results and
figures of the paper entitled "Unfolding the genome: the case study of P.
falciparum".
The repository is organized following the guidelines of "A Quick Guide to
Organizing Computational Biology Projects".
- the *data* lies into the folder called ``data``. To download it
automatically, just go in that folder and type make. Else, make sure to
follow the README in this particular folder to organize the data properly.
- the `script` folder contains all the scripts necessary to reproduce the
figures of the article. Precisely,
- "Figure 3: contact maps of P. falciparum's chr7" can be reproduced by
running `figure_counts.py` in `scripts/visualizing_counts`.
- "Figure 4:" 3D models of P. falciparum's genome architecture"'s code
is contained in `scripts/structures` and can be
reproduced by:
- Inferring the MDS structures for ay-trophozoites and lemieux-B15C2::
python infer_structures_mds.py \
--lengths data/ay2013/trophozoites_10000_raw.bed \
data/ay2013/trophozoites_10000_raw.matrix
python infer_structures_mds.py \
--lengths data/lemieux2013/25kb/B15C2_combined_raw.bed \
data/lemieux2013/25kb/B15C2_combined_raw.matrix
- Inferring the PO structure for the same data sets::
python infer_structures_po.py \
--lengths data/ay2013/trophozoites_10000_raw.bed \
data/ay2013/trophozoites_10000_raw.matrix
python infer_structures_po.py \
--lengths data/lemieux2013/25kb/B15C2_combined_raw.bed \
data/lemieux2013/25kb/B15C2_combined_raw.matrix
- Visualize the structures::
python plot_results.py \
results/lemieux2013/25kb/B15C2_combined_raw_PO_01_structure.txt \
--lengths data/lemieux2013/25kb/A44_combined_raw.bed \
--resolution 25000
python plot_results.py \
results/ay2013/trophozoites_10000_raw_PO_01_structure.txt \
--lengths data/lemieux2013/25kb/B15C2_combined_raw.bed \
--resolution 10000
- "Figure 5: Volume Exclusion Modeling" can be reproduced from the
`scripts/volume_exclusion` folder using the following steps.
- Generating a large number of VE-structures using the following
command::
python plasmo_landmark_without_VRSM.py \
-s troph \
SEED
where SEED is a the seed number. In the paper, I used seeds ranging
from 1 to 5000. Count 1 hour per structures. A slurm script is
avalailable in cluster_scripts. More information in
`scripts/volume_exclusion/README`.
- Then, run::
python create_and_plot_ve_counts.py
- "Figure 6: Stability of structures across the life cycle"'s code is in
`scripts/population_structures`. But first, 1000 structures for
ay-ring, ay-troph, ay-schizont and lemieux-B15C2 need to be computed.
I can provide the set of structures upon demand. Then:
- Create the feature matrices::
python compute_downsampled_distances.py \
structures/ay2013/rings_10000_raw_PO \
--lengths data/ay2013/rings_10000_raw.bed
python compute_downsampled_distances.py \
structures/ay2013/schizonts_10000_raw_PO \
--lengths data/ay2013/rings_10000_raw.bed
python compute_downsampled_distances.py \
structures/ay2013/trophozoites_10000_raw_PO \
--lengths data/ay2013/rings_10000_raw.bed
python compute_downsampled_distances.py \
structures/lemieux2013/25kb/B15C2_combined_raw_PO \
--factor 4 \
--lengths data/lemieux2013/25kb/DCJ_On_combined_raw.bed
- And compute the stochastic PCA::
python clustering_across_stages_downsampled_distances.py -a PO
TODO
- extract useful code from minorswing and add it either to pastis or iced
- make sure anyone can download structures