Other Tools

Table of Contents

• RaptorX
• Conkit
• DeepMSA
• Gremlin
• rMSA

Here, we provide an exploration of the tools which have integrated NEFF calculations into their core functions. We will delve into the primary purposes each of these tools serve and provide an exploration of the various features they offer for executing NEFF computations.

RaptorX

RaptorX, a well-established protein structure prediction method developed since 2012, specializes in predicting protein tertiary and contact structures, particularly for sequences lacking close homologs in the Protein Data Bank [1]. RaptorX includes an integrated Python helper function for NEFF calculation, which is specifically designed for computing asymmetric NEFF. This tool is primarily designed for handling aligned a2m and a3m formats, without the specification of biological sequence alphabets. The code is available on GitHub.

---

Conkit

Conkit is a Python library designed to facilitate the management and manipulation of residue-residue contact prediction data [2]. Among its key functionalities, it provides parsers for different MSA formats, allowing conversion between different MSAs, as well as the the capability to compute NEFF and sequence weights, with a performance boost from Cython for time efficiency. The source code is available on GitHub.
While supporting multiple MSA formats, it does not specify the biological sequence type for NEFF computation. Additionally, when calculating symmetric NEFF, it gives an integer-valued NEFF value without normalization. Furthermore, Conkit can handle gaps in query sequences, but this option is available exclusively in the a3m format, distinguishing it from the a3m-inserts format, which includes gap positions of the query sequence in MSAs.

---

DeepMSA

DeepMSA is an open-source tool designed for constriction of deep and sensitive MSAs [3]. It achieves this by leveraging homologous sequences and alignments derived from a diverse range of whole-genome and metagenome databases, utilizing complementary hidden Markov model algorithms. DeepMSA includes a built-in feature for NEFF computation in C++, which is exclusively available for the aln format. It offers support for both symmetric and asymmetric NEFF calculation for protein alphabets. The source code can be accessed through this link on GtiHub.

---

Gremlin

Gremlin is a method for prediction of residue-residue contacts that uses the power of sequence coevolution and structural context data using a pseudo-likelihood approach. This allows for more precise prediction in protein structures, even when working with a limited set of homologous sequences [4]. Gremlin has been implemented in both Python and C++, and within its code, it includes the capability to compute symmetric NEFF.
When computing NEFF, Gremlin treats all non-standard residues as equivalent to gaps. It can support various biological sequences, including proteins, RNAs, and DNAs. Gremlin offers the feature to exclude gappy positions from the MSA and supports fasta and aln MSA formats. The source code is accessible on GitHub at this link.

---

rMSA

rMSA is a hierarchical pipeline designed for conducting sensitive searches and precise alignments of RNA homologs for a RNA sequence [5]]. rMSA includes a native feature for computing NEFF values specifically customized for RNA sequences. Research in RNA 3D structure prediction and protein-nucleotide structure prediction has employed rMSA to generate MSAs for RNA sequences [6-8].

References

1. Xu, J., et al. "RaptorX: exploiting structure information for protein alignment by statistical inference." Proteins: Structure, Function, and Bioinformatics 80.1 (2012): 62-71.
2. Simkovic, F., et al. "ConKit: a Python interface to contact predictions." Bioinformatics 33.15 (2017): 2209-2211.
3. Zhang, C., et al. "DeepMSA: constructing deep multiple sequence alignment to improve contact prediction and fold-recognition for distant-homology proteins." Bioinformatics 36.7 (2020).
4. Baker, D., et al. "GREMLIN: An Online Server for Protein Structure Prediction." Nucleic Acids Research 41 (2013): W309-W314.
5. Zhang, J., et al. "rMSA: a new method for generating multiple sequence alignments for RNA homologs." Nucleic Acids Research 51.7 (2023): e42.
6. Baek, M., et al. "Accurate prediction of protein-nucleotide complex structures using a deep learning method." Nature 603.7903 (2022): 484-488.
7. Feng, P., et al. "trRosettaRNA: Protein structure prediction with deep learning and co-evolutionary data." Nature Methods 19.5 (2022): 560-565.
8. Zhang, Y., et al. "DeepFoldRNA: accurate RNA secondary structure prediction with deep learning." Nature Communications 13.1 (2022): 1680.

---

For further assistance or inquiries, please contact the developer or create an issue in the GitHub repository.