This program is the second step (downstream) of a protein structure prediction project. This step consists of threading a query sequence on different given templates.
Our project is part of the Meet-U 2018-2019 competition. Meet-U is a collaborative pedagogical and research initiative between several Universities of Paris area. The course is intended to Master students (2nd year) in Bioinformatics. For more details, please refer to http://www.meet-u.org/.
The overall strategy implemented by our team is the following :
- The
.foldrecfile resulting from a profile-profile alignment and generated during the first step (upstream) is parsed to extract relevant informations (query and template sequences, alignment score, etc.). - For each alignment, the query sequence is threaded on the template and a threading score is generated using an energy DOPE matrix.
- Then, the program MODELLER generates a new 3D model by homology (alignment between query and template) and returns a high resolution (bin 0.125Å) DOPE score, which is a statistical potential implemented in MODELLER. This new model is then used to calculate several other scores (secondary structure, solvent accessibility and co-evolution scores).
- Each generated scores are normalized using the min-max scaling method (values between 0 and 1) in order to addition them.
- The scores are stored in
scores.csvand the top N pdb structure are generated.
To improve the structure prediction of the query, several scores have been implemented :
- Alignment score : Resulting from the profile-profile alignment during the first step (Upstream).
- Threading score : Score based on the threading of the query on the template and the energy DOPE matrix.
- Modeller score : Calculating the DOPE statistical potential score of the new model generated by homology modelling with MODELLER based on the alignment of the query and the template.
- Co-evolution score : Co-evolution score measures co-occurence of a pair of amino acid in ortholog sequences using CCMpred program.
- Secondary structure score : Based on the predicted secondary structure and associated confidence score generated by PSIPRED, query and template are compared and a score is generated.
- Solvent accessibility score : DSSP generates a solvent accessibility value for each amino acid of the template and the new template generated by Modeller. Those two values are compared and a solvent accessibility score is generated.
- Blosum score : Comparing amino acid sequence between the query and the template and generating a score based on the blosum62 matrix. This score is not used in the combined score because of its lack of prediction.
- Combined score : The addition of all previously mentionned scores (without the blosum score)
This program is also benchmarked using ROC style plots and Top N information to evaluate the power and the relevance of the different scores. The score results are generated for all queries (foldrec files). Each plot represents the cumulative sum of benchmarks encountered along the ranking (from rank 1 to rank 405) for each calculated scores and for a specific benchmark type ("Family", "Superfamily" or "Fold") which correspond to the degree of similarity with the query.
A top N results table is also generated showing the number of "Family", "Superfamily" and "Fold" benchmark found in the top N ranks.
We used the new data and the new benchmark.list.
git clone https://github.com/meetU-MasterStudents/Fold_U.git
cd Fold_U
Install the few required packages / modules :
pip install -r requirements.txt
MODELLER is also required, and can be installed easily with Conda :
conda install -c salilab modeller
You need to register to get a license key here, and follow instructions during installation to insert license key in the program.
To calculate the solvent accessibility score DSSP is also required:
sudo apt-get install dssp
Or you can download the latest release and install from source.
In order to gain time, we have already run CCMpred with all the multiple alignment files to create the multiple alignment files in .clustal format and the .mat files with the following script :
./script/run_ccmpred.py
fold_u takes in input a foldrec file and a multiple alignment file (clustal format) and the CCMpred result for the studied query. It returns a score.csv file and the top N pdb structures.
The scores.csv and the top 10 pdb structures of the His_biosynth query sequence are stored in results/His_biosynth folder.
./fold_u data/foldrec/His_biosynth.foldrec data/aln/clustal/His_biosynth.clustal\
data/ccmpred/His_biosynth.mat -o results/His_biosynth
./fold_u -h
Usage:
./fold_u FOLDREC CLUSTAL CCMPRED [--nb_pdb NUM] [--output PATH] [--dssp PATH] [--cpu NUM]
[--metafold PATH] [--dope PATH] [--benchmark PATH]
Arguments:
FOLDREC N profile * profile alignment and
their corresponding score.
CLUSTAL Path to the multiple alignment file (clustal format).
CCMPRED Path to the ccmpred result file.
Options:
-h, --help Show this
-p NUM, --nb_pdb NUM Number of pdb to create
[default: 10]
-o PATH, --output PATH Path to the directory containing
the result files (scores and pdb)
[default: ./results]
-a PATH, --dssp PATH Path to the dssp software
binary [default: /usr/bin/mkdssp]
-c NUM, --cpu NUM Number of cpus to use for parallelisation. By default
using all available (0).
[default: 0]
-m PATH, --metafold PATH Path to the metafold.list file
[default: data/metafold.list]
-d PATH, --dope PATH Path to the dope.par file
[default: data/dope.par]
-b PATH, --benchmark PATH Path to the benchmark.list file
[default: data/benchmark.list]
script/benchmarking.py runs the fold_u program for each foldrec if results are not still generated. It returns a results/plots folder containing the generated plots and prints the top N tables in the terminal.
./script/benchmarking.py
./script/benchmarking.py -h
Usage:
./script/benchmarking.py [--selected_score SCORE] [--dssp PATH] [--cpu NUM] [--output PATH]
Options:
-h, --help Show this
-s SCORE, --selected_score SCORE Score for which you wish to see the statistics:
"alignment", "threading", "modeller",
"secondary_structure", "solvent_access"
or "sum_scores",
or all of them at once: "all" [default: all]
-d PATH, --dssp PATH Path to the dssp software
binary [default: /usr/local/bin/mkdssp]
-c NUM, --cpu NUM Number of cpus to use for parallelisation. By default
using all available (0).
[default: 0]
-o PATH, --output PATH Path to the directory containing
the result files (scores and plot)
[default: ./results/plots]
! We used the new data !
Table summarizing the top N results.
Family Superfamily Fold Total
top 5 0/1 1/6 2/13 3/20
0.0 % 16.7 % 15.4 % 15.0 %
-----------------------------------------------------
top 10 0/1 2/6 2/13 4/20
0.0 % 33.3 % 15.4 % 20.0 %
-----------------------------------------------------
top 15 0/1 3/6 2/13 5/20
0.0 % 50.0 % 15.4 % 25.0 %
-----------------------------------------------------
top 20 0/1 3/6 2/13 5/20
0.0 % 50.0 % 15.4 % 25.0 %
-----------------------------------------------------
top 25 0/1 3/6 3/13 6/20
0.0 % 50.0 % 23.1 % 30.0 %
-----------------------------------------------------
top 50 0/1 4/6 6/13 10/20
0.0 % 66.7 % 46.2 % 50.0 %
-----------------------------------------------------
top 75 0/1 4/6 7/13 11/20
0.0 % 66.7 % 53.8 % 55.0 %
-----------------------------------------------------
top 100 0/1 4/6 8/13 12/20
0.0 % 66.7 % 61.5 % 60.0 %
-----------------------------------------------------
top 150 0/1 5/6 8/13 13/20
0.0 % 83.3 % 61.5 % 65.0 %
-----------------------------------------------------
top 200 0/1 5/6 10/13 15/20
0.0 % 83.3 % 76.9 % 75.0 %
-----------------------------------------------------
top 250 0/1 6/6 11/13 17/20
0.0 % 100.0% 84.6 % 85.0 %
-----------------------------------------------------
top 300 0/1 6/6 13/13 19/20
0.0 % 100.0% 100.0% 95.0 %
-----------------------------------------------------
top 350 1/1 6/6 13/13 20/20
100.0% 100.0% 100.0% 100.0%
-----------------------------------------------------
The documentation of our program is generated with Sphinx and and built on Read The Docs.
We are master students in bioinformatics at Paris Diderot University.
Thanks to Maïté Cretin for the nice logo.
Thanks to team 1 and team 2 for generating to us multiple alignment files (data/aln).
This project is licensed under the MIT License.