Awesome

Loop Helix Loop Reshaping

Loop-helix-loop (LHL) units are a type of structual motif that is highly varible in natural proteins and determines protein functions when a LHL unit is a part of a functional site. This repository provides the PyRosetta based method, Loop-helix-loop unit combinatorial sampling (LUCS), to simultaneously sample geometries and lengths of LHL units.

The LUCS method was applied successfully to design de novo protein fold families. The results are describe in this paper:

Pan, X., Thompson, M., Zhang, Y., Lin, L., Fraser, J.S., Kelly, M.J. and Kortemme, T., 2020. Expanding the space of protein geometries by computational design of de novo fold families.

A brief review for using the code can be found here.

Installation

I recommand to create a python virtual environment before installing the package.

virtualenv -p /usr/bin/python3 --system-site-packages venv

Activate the virtual environment and install the package by

pip install -e .

Glossary

Loop-helix-loop (LHL) unit: a segment of a protein that has the loop-helix-loop secondary structure. The two ends of a LHL unit should be anchored on regulary secondary structures (alpha helices or beta strands).

Scaffold: the structure that LHL units are built on. The scaffold is fixed during the LHL modeling process.

Insertion point: a position where LHL units are inserted. An insertion point is represented as a dictionary. For example

{"start":58, "stop":76, "start_ss":"sheet", "stop_ss":"sheet"}

defines an insertion point that starts at sequence position 58 and stops at sequence position 76. The secondary structures that are before and after the LHL units are both beta strands. Note that the start and stop positions are not on the LHL unit, i.e. the start position is the last residue on the preceeding beta strand and the stop position is the first residue on the post beta strand.

Running An Example

This is an example that remodels two loop-helix-loop units in the test_inputs/2lv8_cleaned.pdb structure. The insertion points for the LHL units are defined in the file test_inputs/2lv8_insertion_points.json.

Select linker loops

The first step is selecting linker loops that do not clash with the scaffold.

./run_jobs.py test_select_linkers job_scripts/select_linkers_example.py

This command will create the directory data/test_select_linkers and dump 16 json libraries for selected linker loops at different positions.

Screen loop helix loop units

The second step is screening LHL units that at each insertion points individually. At each insertion point, pairs of linker loops will be inserted and a helix will be introduced to bridge the gap. The built LHL unit will be filtered by the quality of the gap closure, the helix-scaffold contact and the number of buried unsatisfied H-bonds.

./run_jobs.py test_screen_single_insertion_loop_helix_loop_units job_scripts/screen_single_insertion_loop_helix_loop_units_example.py

The command will create the directory data/test_screen_single_insertion_loop_helix_loop_units and dump 2 json libraries for LHL units at the two insertion points. In a production run, you would like to set max_num_success_each_db_pair=None in the script.

Screen compatible loop helix loop units

The last step is to get combinations of LHL units that don't clash against each other.

./run_jobs.py test_screen_compatible_loop_helix_loop_units job_scripts/screen_compatible_loop_helix_loop_units_example.py

The command will create the directory data/test_screen_compatible_loop_helix_loop_units and dump the selected models. For each model, a pdb file of its structure and a json file of its insertion points will be made.

Production Run

Now that you know how to run the example, let's see how to do a production run for a real problem, which often produce tens of thousands of structures. First you need to prepare a pdb file for the scaffold and a json file for the insertion points. I recommand to create a directory under the data directory to store your inputs. Then you need to customize the job scripts. Copy the example scripts to the job_scripts/user directory and edit them for inputs and options. Run a job script by

./run_jobs.py my_output_data_set job_scripts/user/my_script.py

The two LHL screening steps are usually too computationaly expensive to ran on a single computer. You'd better use a cluster. To submit a job to the UCSF Wynton cluster. Just do

./run_jobs.py my_output_data_set job_scripts/user/my_script.py -d SGE -n 100

This command will use 100 CPUs to run the job.