CLAUDIO, the tool for "Cross-linking analysis using distances and overlaps", allows for an in-depth evaluation of structure and sequence information, automating many necessary post-experiment analysis. It downloads protein structures for this, and returns protein-link-specific small-datasets containing structural restraints in CSV-format, and the input dataset extended by its results. These include...
- ... PDB IDs of protein structures searched by BLASTP
- ... Mapping of UniProt protein to structure sequence positions
- ... Structural distances calculated with TopoLink
- ... Information on Homo-signal responses (e.g. overlapping peptide sequences in same-protein cross-links)
- ... Information on possible oligomeric states discovered by SWISS-MODEL homology
- ... Cross-link type estimations
For a quick setup, run the following command with the pip-installation associated with your version of python 3,
while your current working directory is the directory containing this README file (alternatively: replace '.' in the
command with the path to said directory):
pip install .
In order to run CLAUDIO you need to install the following external tools:
- Blast with pdbaa database (see Windows or Unix Manuals, or see this MacOS Manual)
- For Blast, you also have to download the newest pdbaa database. You may do so by navigating into your
Blast installation directory and running the following command:
This download method requires
perl bin/update_blastdb.pl --passive --decompress pdbaaperlto be executable from your commandline, which is not installed by default on Windows and MacOS systems. You may install it with this Windows-Manual or MacOS-Manual, or manually download the database here. - Hint: If you followed the Installation Manual (linked up top) to-the-letter, you may have added the environmental
variable
$BLASTDBto your paths. If so, you can use this variable instead of the full path in the input parameters of CLAUDIO.
- For Blast, you also have to download the newest pdbaa database. You may do so by navigating into your
Blast installation directory and running the following command:
- TopoLink (see Installation Manual)
- Topolink possesses a standalone executable for Windows 10 systems (or higher), which can be downloaded directly here.
To ease the execution of CLAUDIO you may want to ensure that these tools can be executed directly from your terminal
(e.g. add their respective bin directories to the Path variable of your OS). Otherwise, you may specify the
location of the bin directories in the input parameters.
CLAUDIO calls upon the API of a number of bioinformatic online databases (UniProt
[3], RCSB
[4], AlphaFold
[5], and SWISS-MODEL
[6]) during its computations. This means it cannot be
run offline.
It is furthermore recommended having a stable internet connection, as otherwise certain API calls may not be answered or
lead to empty results. This of course, also necessitates the database server's side to be running properly as well. If
errors or suspicious inconsistencies in the results persist due to this, you may want to try again later.
In addition to the aforementioned online databases, CLAUDIO accesses the SIFTS database
[7,8]. The file in question can be found
here.
We also recommend updating this file from time to time (download here)
in order to keep up its efficiency, though this is not a necessity (last updated: 26.09.2023).
This tool is written in and has to be run with python 3 (last tested v3.11).
It has the following requirements:
- biopython 1.79
- click 8.1.3
- matplotlib 3.6.3
- pandas 1.5.3
- requests 2.28.2
The packages may be installed all at once with the file requirements.txt:
pip install -r requirements.txt
or individually:
pip install biopython==1.79
pip install click==8.1.3
pip install matplotlib==3.6.3
pip install pandas==1.5.3
pip install requests==2.28.2
Note: Both approaches need to refer to the pip-installer associated to the python installation, that will be used to run the tool.
CLAUDIO consists of a total of 4 modules. Each module can be run independently as long as appropriate inputs are delivered. For details on how to run the modules individually see their respective README-files.
- Module 01 - Unique protein (pair) listing tool
- Module 02 - Structural distance analysis tool
- Module 03 - Overlapping peptide sequence analysis tool
- Module 04 - XL-type evaluation
For details on how to run the full pipeline continue below.
> claudio [-i <filepath>] [-it <diretorypath>] [-o <directorypath/"">] [-p <"comma-separated str">] [-bl <directorypath/None>] [-bldb <directorypath>] [-tl <directorypath>] [-x <comma-separated str>] [-lmin <float>] [-lmax <float>] [-t <"blastp">] [-e <float] [-qi <float>] [-cv <float>] [-r <float>] [-rt <True/False>] [-pc <float>] [-s <True/False>] [-v <int>] [-es <float>] [-dm <float>] [-ct <float>] [-c <filepath>]
-i, --input-filepath, path to inputfile, default="test/sample_data_random.csv"
-it, --input-temppath, path to directory for temporary files, default=None
-o, --output-directory, output directory for produced csv-files, default="test/out/sample"
-p, --projections, comma-separated position-sensitive list that names the column names of the users dataset
containing the necessary information for the tool. The column names should contain and
should be given in the following order: crosslinked peptide_a, crosslinked peptide_b,
crosslinked residue position_a, crosslinked residue position_b, position of cross-linked
residue in peptide_a, position of cross-linked residue in peptide_b, UniProt ID of
protein belonging to peptide_a, UniProt ID of protein belonging to peptide_b.
Note: The positions of the crosslinked residue in the peptides are information only
accessed, if the given full sequence positions do not match into the retrieved UniProt
sequence. If the positions are confirmed you may simply create two substitute columns
for the positions in the peptides instead and leave them empty.
default="peptide1,peptide2,position1,position2,k_pos1,k_pos2,entry1,entry2"
-bl, --blast-bin, binary directory in blast installation, or None if binary directory has been added to
PATH variable (e.g. if blast can be called from anywhere), default=None
-bldb, --blast-db, database directory for blast installation, default="$BLASTDB"
-tl, --topolink-bin, binary directory in topolink installation, or None if binary directory has been added to
PATH variable (e.g. if topolink can be called from anywhere), default=None
-x, --xl-residues, comma-separated one-letter-code residues, optional: add two ':' after the
one-letter-code symbol of the residue in order to specify full sequence position
(either 1 for start, or -1 for end position) and/or the atom used for the distance
computation (allowed: "N", "CA", "C", "O", "CB"), default="K,M:N:1"
-lmin, --linker-minimum, float value used as minimal crosslinker range in angstrom, default=5.0
-lmax, --linker-maximum, float value used as maximal crosslinker range in angstrom, default=35.0
-t, --search-tool, always set to "blastp" (as of this version), specifying the tool which should be used for pdb
search, default="blastp"
-e, --e-value, e-value used in structure search, default=1e-5
-qi, --query-id, query identity used in structure search, default=90.0
-cv, --coverage, coverage used in structure search, default=50.0
-r, --res-cutoff, float value used as cutoff in angstrom for resolution of structure files, default=4
-rt, --read-temps, if the tool has been run before with the same input a temporary file was saved, which
can be used to skip some of the steps, default=False
-pc, --plddt-cutoff, float value used as cutoff for alphafold structure prediction confidences (plddt),
default=70.0
-s, --compute-scoring, boolean, for whether experimental scoring and resulting XL-type evluations should be
computed and appended to result dataset, default=False
-v, --verbose-level, verbose level value, default=3:
0: no outputs at all will be written to the commandline
1: write tool inits and passed time
2: write progressbars (where implemented)
3: write alignments during data processing, and write extra information on process results
4: write alignments during uniprot to pdb position translation
5: write verfications during uniprot to pdb position translation
-es, --euclidean-strictness, float value substracted from the linker ranges for the euclidean distance scoring
(minimum will not go below 0), default=5.0
-dm, --distance-maximum, maximal distance value that seems realistic, if surpassed the distance will be set to
this value during the confidence scoring, to ensure its consistency, default=50.0
-ct, --cutoff, float value used as confidence score cutoff, if surpassed, the linker type will be set
to inter, default=0.0
-c, --config, filepath to configuration file containing all input parameters, default=''
This tool requires a CSV-file containing multiple observed cross-linking interactions. Two columns have to contain uniprot ids for each interacting residue, two columns have the observed peptides for each interacting residue and two columns the crosslinked residue's position within the full sequence (alternative: fill these with Nans, but add two columns with the residue's positions in the respective peptides).
- The input file should then be specified as such with the "-i / --input-filepath"-parameter.
- If you intend to run CLAUDIO on separate datasets simultaneously, or want to split your dataset into smaller ones, you have to specify the parameter "-it / --input-temppath". CLAUDIO generates multiple temporary files during its computation, most of which are in- or outputs of the third-party tools used. If parallel executions of CLAUDIO are run with the same tempfile path, conflicts may be caused disrupting or falsifying some results. Thereby make sure to specify different paths here, if you run the tool in parallel.
- It is important to customize the parameter "-p / --projections". This parameter requires a comma separated list as input, which maps the column names of your dataset to the ones used in the tool.
- You need to specify the paths to the local external tool installations (BLASTP and TopoLink). The parameters for this are "-bl / --blast-bin" for the binary directory of blast, "-bldb / --blast-db" for the database directory containing the pdbaa database files, and "-tl / --topolink-bin" for the binary directory of TopoLink.
- Make sure you customize the settings pertaining the cross-linking experiments specifications, e.g.
"-x / --xl-residues" for the aminoacids which may be cross-linked, and both "-lmin / --linker-minimum" and
"-lmax / --linker-maximum" for the cross-linker's range capability. Besides this specify the amino acids the used cross-
linker is able to bind to. For this specify a tiny comma-separated list of possibly crosslinked residues, as
one-letter-code symbols. Optionally you may add two colon-symbols, if they wish to specify the position of the
residue in the sequence and/or the atom in the residue used for the distance computation. After the first colon-symbol
they may place the atom type for the distance computation, e.g. 'CB', 'CA', or 'N', etc. . If no value is set here 'CB'
will be used by default. After the second colon-symbol they may place one of three position values: 0 if the residue
may occur anywhere in the protein, 1 if the residue has to be at the N-terminus, or -1 if the residue has to be at the
C-terminus (ex.: use "K:CB:0" to calculate the cross-link distance between lysin C-beta-atoms at any position in the
chain). If you want to specify multiple possible positions for the same symbol you have to add them
individually (ex.: "M::1,M::-1" for methionine at either the beginning or end of the chain). By default, 0 is set here,
e.g. the residue may be placed anywhere.
Note: If the position or the atom type is specified there have to be two colon-symbols (ex.: "K,M:1" will not be accepted as input). You can leave the respective specification empty though, if you wish to use the default here (ex.: "K::1" is equal to "K:CB:1", "M:N:0" is equal to "M:N:", "K:CB:0" is equal to "K::" and also to just "K").
With this the relevant settings are defined. You may choose to further specify the structure searche's settings in terms of coverage, sequence identity, and e-value, as well as the resolution cutoff during the structure selection, or even the advanced settings.
You may see examples for all parameters in the example configuration-file, which also serves as an alternative to supply all parameter inputs.
This tool returns all the outputs listed in the modules (see module01, module02, module03, module04).
Note: All CSV-file outputs pertaining the input dataset are summarized into a single one (marked with '_final.csv'-extension), e.g. the output CSV-file of module01 ending with '.sqcs', of module02 ending with '.sqcs_structdi.csv', and of module03 ending with '.sqcs_ops.csv' will be summarized here.
CLAUDIO (in full) can be run like this:
- with a configuration file with all parameters (this will run the test dataset, when using the default config.txt, with all non-described parameters being filled with default values)
claudio -c config/config.txt
- with a configuration file and a few overwriting CLI parameters (this will run the larger benchmark dataset)
claudio -c config/config.txt -i benchmark_data.csv -o test/out/benchmark
- with mostly default parameters and a few set as CLI parameters
claudio -i /home/user/me/docs/xl_dataset.csv -o /home/user/me/docs/claudio_outputs
Also, you may return all CLI parameter options on the terminal like this:
claudio --help
- Alexander Röhl
- Hadeer Elhabashy
- Eugen Netz
- [1] Ferrari, Allan JR, et al. "TopoLink: evaluation of structural models using chemical crosslinking distance constraints." Bioinformatics 35.17 (2019): 3169-3170.
- [2] Altschul, Stephen F., et al. "Basic local alignment search tool." Journal of molecular biology 215.3 (1990): 403-410.
- [3] UniProt Consortium. "UniProt: a worldwide hub of protein knowledge." Nucleic acids research 47.D1 (2019): D506-D515.
- [4] Kouranov, Andrei, et al. "The RCSB PDB information portal for structural genomics." Nucleic acids research 34.suppl_1 (2006): D302-D305.
- [5] David, Alessia, et al. "The AlphaFold database of protein structures: a biologist’s guide." Journal of molecular biology 434.2 (2022): 167336.
- [6] Schwede, Torsten, et al. "SWISS-MODEL: an automated protein homology-modeling server." Nucleic acids research 31.13 (2003): 3381-3385.
- [7] Dana, Jose M., et al. "SIFTS: updated Structure Integration with Function, Taxonomy and Sequences resource allows 40-fold increase in coverage of structure-based annotations for proteins." Nucleic acids research 47.D1 (2019): D482-D489.
- [8] Velankar, Sameer, et al. "SIFTS: structure integration with function, taxonomy and sequences resource." Nucleic acids research 41.D1 (2012): D483-D489.