Complet-Plus: A Computationally Scalable Method to Improve Completeness of Large-Scale Protein Sequence Clustering

Complet+ is a post-processing tool that merges clustering results using MMseqs2's search module. It is intended for useage with MMseqs2 clustering results, but can be run on any clustering results that is in a similar format.

Drexel University EESI Lab, 2022

Maintainer: Rachel Nguyen, rtn28 at dragons dot drexel dot edu

Owner: Gail Rosen, gailr at ece dot drexel dot edu

Dependencies

Complet+ was developed with the following software:

• python=3.9
• MMseqs2=Release 13-45111

Data requirement

1. clusterResults.tsv

The clustering results file format requires that each cluster have a representative sequence.

Please note that the actual clusterResults.tsv file itself should not have headers.

Representative sequence	Sequence
SEQUENCE_1	SEQUENCE_1
SEQUENCE_1	SEQUENCE_2
SEQUENCE_1	SEQUENCE_3
SEQUENCE_4	SEQUENCE_4
SEQUENCE_4	SEQUENCE_5
...	...

2. sequences.fasta example

>SEQUENCE_1
rsiwskaggsaeeigaealgrmle
>SEQUENCE_2
tsadkshvrsiwskaggsaeeigaealgrmlesf
>SEQUENCE_3
wskaggsaeeigaealgrmle
...

Output format

There are 3 columns: Old Cluster ID, Sequence ID, Complet+ Cluster ID

Tutorial

completplus.sh

Complet+ is run via the command line using the script completplus.sh.

If you are using Complet+ via the singularity container, the script's directory is already added to the PATH variable so you can simply run it as follows:

completplus <i:clusterResults.tsv> <i:sequences.fasta> <o:newClusteringResults.tsv> [options]

The arguments for the script are: the clustering results file that the user wishes to run Complet+ on, the FASTA file of sequences, and the name of the new clustering results file that Complet+ will make.

The user may also specify the options they wish to run MMseqs2's search with, as a string. This string is passed straight to the MMseqs2 search call, so any options that are availble to MMseqs2 search are available for use. If the user wishes to increase the amount of merging Complet+ does, they can increase the e-value threshold from its default value of [-e 1.000E-03].

For example, let's say we have a clustering results file called defaultClusters.tsv, a FASTA file called allSeqs.fasta, and we would like the resultant file to be called completClusters.tsv. To run Complet+ with using a MMseqs2 search sensitivity of 7.5 and an e-value threshold of 0.1, the command would look like the following:

completplus defaultClusters.tsv allSeqs.fasta completClusters.tsv "-s 7.5 -e 0.1"

find_RHs.py

The Python script run by completplus.sh to filter the sequence alingment down to the reciprocal hits. Not intended for user use.

relabel_seqs.py

The Python script run by completplus.sh to relabel the sequences using the reciprocal hits from the sequence alignment. Not intended for user use.

Running the Docker in Singularity (Example of using the Docker)

singularity pull docker://eesilab/complet-plus:0.1

To run example (where you want to output tempdir and output in current directory): singularity exec -B $PWD:/data completplus_amd.sif bash completplus.sh -c /opt/complet-plus-scripts/example_input_files/step-0.tsv -s /opt/complet-plus-scripts/example_input_files/step-0.fasta -o /data/step-1.tsv