# NCBI Conserved Domain Viral Protein: Step-by-Step Guide By the end of this guide you will be able to take any viral [protein sequence](/blog/guides/protein-sequence), submit it to the NCBI Conserved Domain Database (CDD), identify its functional domains, interpret the statistical output, and use those results for vaccine design, drug targeting, phylogenetic placement, or functional annotation. You will learn both the web-based workflow and the command-line alternative using rpsblast. No prior experience with CDD is required, though basic familiarity with FASTA format and BLAST concepts helps. ## What Is the NCBI Conserved Domain Database The Conserved Domain Database is a curated collection of [multiple sequence alignment](/blog/guides/multiple-sequence-alignment) models that represent conserved protein domains. A domain is a segment of a protein that folds and functions independently, often carrying a specific biochemical activity such as binding, catalysis, or structural support. CDD is maintained as part of the broader suite of NCBI analysis and retrieval resources, which include Entrez, BLAST, Gene, Taxonomy Browser, and the Protein Clusters database [1]. CDD imports domain models from several partner databases. It includes curated domains produced by NCBI annotators (accessions beginning with `cd`), Pfam domains (accessions beginning with `pfam`), SMART domains (accessions beginning with `smart`), COG orthologous groups, and TIGRFAM models. Each domain model carries a position-specific scoring matrix (PSSM) built from the aligned sequences of that domain family. When you submit a query protein, the CD-Search tool runs a reverse position-specific BLAST (RPS-BLAST) against these PSSMs to find regions of your protein that match known domains. The database organizes hits hierarchically. At the broadest level are superfamilies, which group domains that share a common evolutionary origin even if sequence similarity is low. Below superfamilies are families and subfamilies with progressively tighter sequence and functional constraints. A hit to a specific subfamily model carries more functional information than a hit to a broad superfamily model. ## Why Conserved Domains Matter in Viral Proteins Viral genomes are small and evolve rapidly, yet the domains that perform essential functions, such as receptor binding, genome replication, and capsid assembly, resist change because mutations in them are lethal or severely deleterious. This conservation makes domains reliable anchors for annotation and comparison across viral isolates, species, and even families. Several published studies illustrate the range of applications. In the search for Epstein-Barr virus vaccine targets, researchers retrieved gp350/220 glycoprotein sequences from NCBI, aligned them to identify the conserved region responsible for CR2 receptor binding, and then mapped B-cell epitopes within that conserved block [2]. For SARS-CoV-2, analysis of 57,311 [spike protein](/blog/guides/spike-protein) sequences from the NCBI Virus database showed that the receptor-binding domain and three surface N-glycosylation sites were 99.67 percent conserved across variants [3]. A separate study found that the S1 subunit of the spike protein was 99.70 percent conserved, with the receptor-binding domain (residues 319 to 541) showing 100 percent identity among the sequences examined [4]. In bovine leukemia virus, six conserved segments in the envelope gp51 protein showed amino acid conservation between 98.8 and 99.8 percent across 256 global isolates retrieved from NCBI [5]. Beyond epitope mapping, conserved domains serve as phylogenetic markers. The CRESS domain within the replication-associated protein of single-stranded DNA viruses resolves phylogenetic groups consistently with full-length protein trees, making it a compact alternative marker for molecular characterization [6]. At the largest scale, conserved ancestral genes shared across nucleocytoplasmic large DNA viruses (NCLDV) form the basis for reconstructing the evolutionary history of giant viruses [7], [8]. Domain analysis also reveals unexpected evolutionary events. The F12 protein of chordopoxviruses, long implicated in virion morphogenesis, was shown through computational domain analysis to be a derived, inactivated [DNA polymerase](/blog/guides/dna-polymerase) of probable bacteriophage origin [9]. In DNA virus transcription regulators, extensive domain shuffling has been documented, with domains such as KilA-N and Bro-N appearing in multiple combinations across poxviruses, baculoviruses, and other large DNA viruses [10]. ## Prerequisites ![Decision path asking how many proteins you have and whether you need automation, leading to web CD-Search, Batch CD-Search or rpsblast.](/article-images/illustrations/ncbi-conserved-domain-viral-protein-cdd-triage.webp) Before starting, gather the following: 1. A viral protein sequence in FASTA format. You can retrieve one from the [NCBI Protein database](/blog/guides/ncbi-protein-database) by searching for the virus name and protein of interest, then selecting "FASTA" from the display options. For batch work, the NCBI Virus database allows bulk downloads filtered by taxon, host, and geographic region [11], [12]. 2. A web browser for the graphical CD-Search interface. 3. Optional: a Linux or macOS terminal with the NCBI BLAST+ toolkit installed for command-line searches. Install it with `conda install -c bioconda blast` or download from the NCBI FTP site at `ftp.ncbi.nlm.nih.gov/blast/executables/blast+/LATEST/`. 4. Optional: R or Python for downstream visualization of domain architectures. ## Step 1: Retrieve a Viral Protein Sequence from NCBI If you already have a sequence, skip to Step 2. To retrieve one from NCBI: 1. Go to `https://www.ncbi.nlm.nih.gov/protein/`. 2. Type a query such as `SARS-CoV-2 spike glycoprotein[Title] AND srcdb_refseq[PROP]` to limit to RefSeq records. 3. Select the record of interest. 4. From the "Send to" menu, choose "File", set the format to "FASTA", and click "Create File". The downloaded file will look like this: ``` >YP_009724390.1 surface glycoprotein [Severe acute respiratory syndrome coronavirus 2] MFVFLVLLPLVSSQCVNLTTRTQLPPAYTNSFTRGVYYPDKVFRSSVLHSTQDLFLPFFS NVTWFHAIHVSGTNGTKRFDNPVLPFNDGVYFASTEKSNIIRGWIFGTTLDSKTQSLLIV NNATNVVIKVCEFQFCNDPFLGVYYHKNNKSWMESEFRVYSSANNCTFEYVSQPFLMDLE GKQGNFKNLREFVFKNIDGYFKIYSKHTPINLVRDLPQGFSALEPLVDPIGITINITRFQ ... ``` For large-scale studies, the NCBI Virus database at `https://www.ncbi.nlm.nih.gov/labs/virus/vssi/` provides a graphical interface to filter sequences by virus taxonomy, host, collection date, and geography, then export them as FASTA [11], [12]. ## Step 2: Run a Single-Sequence CD-Search on the Web This is the fastest way to annotate one protein. 1. Navigate to `https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi`. 2. Paste your FASTA-formatted protein sequence into the text box, or click "Choose File" to upload a FASTA file. 3. Under "Search options", leave the E-value threshold at the default of 0.01 unless you have a reason to change it. 4. Click "Submit". The search typically completes in under a minute for a single protein. The results page displays a graphical domain architecture with colored boxes representing detected domains positioned along the protein length. Below the graphic, a table lists each hit with its domain accession, name, E-value, bit score, alignment start and end coordinates, and whether the hit is specific or superfamily-level. A typical result for a coronavirus spike protein query would show a large hit to a viral surface glycoprotein domain model spanning most of the ectodomain region, with the receptor-binding subdomain highlighted as a narrower child model within the same superfamily hierarchy. ## Step 3: Run a Batch CD-Search for Multiple Proteins When annotating dozens to thousands of viral proteins, use the Batch CD-Search web tool. 1. Prepare a multi-FASTA file with one protein per entry. Each header must begin with `>` and contain a unique identifier: ``` >BLV_gp51_K02120 MPGGLSWVSTVSSGDLGVAHWTPAQEVIQGVSPEPQARL... >EBV_gp350_AY205202 MPPRRPPPPGPGPVLGIGPVLGLVLAAPWPPGVLLSS... >SARS2_spike_YP009724390 MFVFLVLLPLVSSQCVNLTTRTQLPPAYTNSFTRGVY... ``` 2. Go to `https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi`. 3. Upload your FASTA file or paste the sequences. 4. Select the search mode. "Live search" runs against the current CDD release. "Precomputed search" checks whether your protein already has results stored from a prior run, which is faster for common proteins. 5. Set the E-value threshold. The default of 0.01 is appropriate for most viral protein analyses. 6. Choose an output format. "Interactive" gives a web page. "Tabular" gives a plain-text file suitable for parsing in R or Python. 7. Click "Submit Query". For a batch of 50 to 500 proteins, a live search typically finishes in minutes. The tabular output has columns for query ID, domain accession, domain name, E-value, bit score, alignment coordinates, and completeness of the domain match. ## Step 4: Run rpsblast from the Command Line For integration into automated pipelines or for very large datasets, run RPS-BLAST locally. First, download and format the CDD database: ```bash # Create a working directory mkdir cdd_db && cd cdd_db # Download the CDD database (approximately 1 GB compressed) wget ftp://ftp.ncbi.nlm.nih.gov/mmdb/cdd/CDD_NCBI.tar.gz # Extract tar -xzf CDD_NCBI.tar.gz ``` The extracted directory contains pre-formatted database files with names like `Cdd`, `Cdd.pal`, and volume suffixes. Place them in a directory included in your `BLASTDB` environment variable, or run rpsblast from that directory. Run the search: ```bash rpsblast \ -query viral_proteins.fasta \ -db Cdd \ -evalue 0.01 \ -outfmt "6 qseqid sseqid evalue bitscore pident length qstart qend sstart send" \ -max_target_seqs 50 \ -num_threads 4 \ -out cdd_results.tsv ``` Expected tabular output (first few lines): ``` SARS2_spike_YP009724390 cd21447 3.2e-118 345.1 41.2 685 1 685 1 685 SARS2_spike_YP009724390 pfam04551 1.8e-95 287.6 38.7 610 1 610 1 610 BLV_gp51_K02120 cd00204 4.1e-52 168.3 30.1 245 12 256 1 245 EBV_gp350_AY205202 pfam00595 2.7e-38 128.9 25.4 180 45 224 1 180 ``` Each line represents one domain hit. The `sseqid` column gives the CDD domain accession, which you can look up at `https://www.ncbi.nlm.nih.gov/Structure/cdd/cdds.shtml` for full annotation, alignment, and taxonomy information. The following table summarizes the key parameters: | Parameter | Default | What it controls | When to change it | |--|--|--|--| | E-value | 0.01 | [Statistical significance](/blog/guides/statistical-significance) threshold | Lower to 0.001 for high-confidence hits only, raise to 1.0 for remote homologs | | max_target_seqs | 500 | Maximum domain hits reported per query | Lower to speed output, raise if your protein has many repeated domains | | num_threads | 1 | CPU cores used | Increase on multi-core systems for large batches | | outfmt | Varies | Output columns | Use tabular (6) for parsing, pair (0) for human reading | ## Step 5: Interpret the Results ### E-value and Bit Score The E-value is the expected number of random hits with a score at least as good as the observed one, given the database size. An E-value of 0.001 means you would expect one hit this good by chance in 1,000 searches of a database this size. For viral protein domain assignment, E-values below 0.001 are generally reliable. Values between 0.01 and 0.1 warrant manual inspection of the alignment. The bit score is a normalized measure of alignment quality that is independent of database size. Higher bit scores indicate stronger matches. A bit score above 100 typically indicates a highly significant match. ### Specific vs. Superfamily Hits CDD reports two categories of hits. A "specific" hit means your protein matched a domain model at the family or subfamily level, giving you a precise functional annotation. A "superfamily" hit means the match was detectable only at the broader structural level, suggesting shared ancestry but not necessarily the same function. For example, a hit to the P-loop NTP hydrolase superfamily tells you the protein likely binds nucleotides, but a specific hit to the DEXDc box helicase family within that superfamily tells you it is a DEAD-box helicase. In a study of Japanese eel RIG-I, NCBI CDD analysis identified two adjacent CARD domains, a DEXDc domain, a HELICc domain, and a C-terminal regulatory domain, each carrying distinct functional annotations [13]. ### Domain Architecture The graphical output shows domains as colored boxes along the protein. Overlapping boxes at different levels of the hierarchy represent the same region annotated at different specificity levels. Gaps between domains indicate regions with no detectable similarity to known models, which may represent flexible linkers, novel domains, or disordered regions. For viral polyproteins, such as those of picornaviruses, the domain architecture reveals the positions of individual functional units within the single long chain. In the leafhopper-infecting HoCV-1, analysis of the capsid polyprotein identified the order of subunits CP2, CP4, CP3, and CP1, with cleavage boundaries determined by alignment of conserved motifs at domain junctions [14]. ### Multiple Domain Hits on One Protein Large viral proteins often contain multiple domains. The RSV L protein, for example, carries the RNA-dependent [RNA polymerase](/knowledge/bioinformatics/rna-polymerase-structure-transcription-mechanisms) domain plus a polyribonucleotidyltransferase (PRNTase) domain. A structural analysis of 28,140 RSV L sequences from NCBI Virus and GISAID found low variability within the PRNTase domain and near-complete conservation of the non-nucleoside inhibitor binding pocket, supporting it as a stable drug target [15]. ## Step 6: Apply the Results ### Vaccine and Epitope Design Conserved domains are prime candidates for vaccine targets because they are present across viral variants and less likely to escape immune recognition. After identifying conserved regions through CDD, researchers typically predict T-cell and B-cell epitopes within those regions using tools such as NetMHC, IEDB, or Discotope. In the EBV gp350/220 study, the conserved domain responsible for CR2 binding was mapped and then screened for epitopes using Discotope and BepiPred methods [2]. In a separate approach targeting HHLA2 and MAGEB5 protein variants associated with viral diseases, domain analysis using NCBI CDD informed the selection of antigenic regions for peptide-MHC binding predictions [16]. For SARS-CoV-2, the high conservation of the receptor-binding motif within the RBD made it a template for pharmacophore modeling. One study retrieved 483 RBD sequences from NCBI, aligned them, confirmed that the receptor-binding motif residues interacting with ACE2 were conserved in all sequences, and used that conserved motif to design a pharmacophore for virtual screening of lead compounds [17]. ### Drug Target Identification Domains that are conserved across viral isolates and structurally stable make good antiviral targets. The SARS-CoV-2 spike N-glycosylation sites, found to be 99.67 percent conserved across 57,311 sequences, were proposed as targets for antibody-drug conjugates designed to impair viral infectivity [3]. The RSV L protein PRNTase domain, with its conserved inhibitor binding pocket across 28,140 sequences, represents another example of a conserved domain supporting antiviral drug development [15]. ### Phylogenetic Analysis Conserved domains can substitute for full-length proteins in phylogenetic reconstruction when full-length sequences are unavailable or too divergent. The CRESS domain, approximately 300 base pairs within the begomovirus Rep protein, resolved the same five phylogenetic groups as the full-length Rep protein across 278 sequences from NCBI, with tanglegram analysis confirming one-to-one correspondence between the two trees [6]. At a broader scale, phylogenetic analysis of conserved ancestral NCLDV genes demonstrated that the three groups of giant viruses (Mimiviruses, Pithoviruses, and Pandoraviruses) each evolved independently from smaller DNA viruses within the Megavirales, rather than from a fourth domain of cellular life [8]. ### Functional Annotation of Novel Viruses When a new virus is sequenced, CDD provides the first pass of functional annotation. Domains detected in the proteome suggest what the virus can do. For newly discovered insect viruses such as HoCV-1, identification of capsid protein domains and their cleavage boundaries supported classification and comparison with related picorna-like viruses [14]. ## Common Mistakes and Limitations ### Submitting Nucleotide Sequences to CD-Search CD-Search works on protein sequences. If you paste a nucleotide FASTA, the tool will either return no hits or produce meaningless results. Translate your coding sequence first using ORFfinder (`https://www.ncbi.nlm.nih.gov/orffinder/`) or the `translate` function in Biopython. ### Ignoring the Domain Hierarchy A superfamily-level hit does not give the same functional information as a specific hit. Reporting "P-loop NTP hydrolase superfamily" without noting that the protein is specifically a DEXDc helicase loses actionable detail. Always check whether a more specific child model exists below the superfamily hit in the CDD hierarchy. ### Trusting Low E-values Without Checking Coverage A low E-value with poor domain coverage (for example, only 30 of 200 domain positions aligned) may indicate a partial domain, a fragment, or a spurious match. Check the alignment start and end coordinates against the domain model length. A hit covering less than half the domain model should be treated with caution. ### Using a Single Sequence for Conservation Claims Claims about conservation require alignment of multiple sequences across relevant diversity. A single protein submitted to CDD tells you what domains it has, not how conserved those domains are across variants. For conservation studies, retrieve all available sequences from NCBI Virus or GenBank, align them, and measure identity at each position. Studies that report conservation percentages, such as the 99.67 percent conservation of SARS-CoV-2 spike glycosylation sites across 57,311 sequences [3], or the 98.8 to 99.8 percent conservation of BLV gp51 segments across 256 isolates [5], all started with multi-sequence datasets. ### Assuming Absence of a Domain Means Absence of Function CDD detects similarity to known domain models. A region with no detectable domain may still be functional, especially in viruses where rapid evolution can obscure homology. The SWIRM domain, now recognized in chromosomal proteins, was itself a novel discovery from computational sequence-profile analysis before it entered any domain database [18]. Novel domains in viral proteins continue to be described through similar approaches [10], [19]. ### Confusing CDD with BLASTp ![Side-by-side comparison of CDD profile searches and BLASTp on what is searched, sensitivity to remote homologs and what a null result means.](/article-images/illustrations/ncbi-conserved-domain-viral-protein-cdd-vs-blastp.webp) BLASTp compares your protein to individual protein sequences. CDD compares your protein to pre-built domain profiles (multiple sequence alignments converted to PSSMs). Profile-based searches are more sensitive for detecting remote homologs because they use the conservation pattern across an entire family, not just pairwise similarity to one sequence. If BLASTp returns no significant hits, CDD may still detect a domain. ### Overlooking Post-Translational Modifications CDD identifies protein domains based on sequence. It does not predict glycosylation sites, phosphorylation sites, or other modifications. For SARS-CoV-2 spike protein, the N-glycosylation sites that shield the virus from immune recognition were identified through separate glycosylation analysis tools, not through CDD alone [3]. ### Not Checking for Domain Architecture Changes Some viral proteins undergo domain gain, loss, or shuffling during evolution. A protein that normally contains two domains may appear with only one in some isolates due to truncation or frameshift. The integration of an anellovirus genome into a human chromosome produced a truncated ORF1 capsid protein that retained the jelly roll and P-domain features but lacked the C-terminal domain [20]. Always inspect the full domain architecture, not just the list of hits. ## Tips for Better Results 1. **Use RefSeq protein records when available.** They are curated and annotated, reducing errors from sequencing artifacts or annotation inconsistencies. 2. **Run both specific and superfamily searches.** If CDD returns only superfamily hits, try searching against the curated (NCBI-generated) domain subset separately from the full database including imported models. 3. **Combine CDD with structural data.** For proteins with known or predicted structures, map domain boundaries onto the 3D structure using PyMOL or ChimeraX. Studies of coronavirus spike protein RBDs combined sequence analysis from NCBI with crystal structures from the [Protein Data Bank](/blog/guides/protein-data-bank) to visualize electrostatic changes between SARS-CoV, MERS-CoV, and SARS-CoV-2 [11]. 4. **Check for domain co-occurrence.** The Conserved Domain Architecture Retrieval Tool (CDART), accessible from CDD results, finds proteins with similar domain architectures. This helps identify functional partners or homologs in other viruses. 5. **Validate domain boundaries experimentally.** Domain predictions from CDD provide boundaries based on sequence alignment. For functional studies such as epitope mapping or drug design, confirm that the predicted boundaries correspond to structural or biochemical boundaries. 6. **Export domain coordinates for downstream analysis.** The tabular output from batch CD-Search or rpsblast includes alignment start and end coordinates. Parse these with Python or R to generate domain architecture diagrams, calculate domain-level conservation, or map mutations onto domain regions. ## Frequently Asked Questions ### Can I use CDD to analyze nucleotide sequences directly? No. CDD accepts protein sequences only. If you have a [nucleotide sequence](/knowledge/molecular-biology/nucleotide-sequence), translate it to protein first using ORFfinder at `https://www.ncbi.nlm.nih.gov/orffinder/` or a local tool such as EMBOSS transeq, then submit the translated protein to CD-Search. ### What is the difference between CDD and Pfam? CDD is a database that includes Pfam models among its sources, alongside NCBI-curated domains, SMART, COG, and TIGRFAM. When you search CDD, you are searching against all of these model sets simultaneously. Pfam alone is a separate database maintained by EMBL-EBI with its own HMM-based search tool. CDD adds NCBI curation, hierarchical organization, and integration with other NCBI resources. ### What E-value threshold should I use for viral proteins? The default of 0.01 works well for most viral protein domain assignments. For high-confidence annotations in publications, use a stricter threshold such as 0.001. For detecting remote homologs in highly divergent viruses, you can relax to 0.1, but manually inspect those alignments to confirm biological plausibility. ### How do I search thousands of viral protein sequences at once? Use the Batch CD-Search web tool at `https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi` for up to a few thousand sequences, or run rpsblast locally with the CDD database for larger datasets. The command-line approach handles tens of thousands of sequences efficiently with multi-threading. ### What does it mean when CDD reports only a superfamily hit and no specific hit? It means your protein matched a domain at the broad structural level but not closely enough to any specific family model within that superfamily. The protein likely contains a divergent member of that domain family. You can explore the superfamily alignment in CDD to see which residues are conserved and infer possible function from the superfamily annotation. ### Can CDD identify completely novel domains? No. CDD detects similarity to known domain models. A region with no CDD hit may contain a novel domain. Discovering novel domains requires de novo computational methods such as profile-profile comparisons, structural prediction with tools like AlphaFold, or remote homology detection with HHpred, which compares your sequence against profile HMM libraries at greater sensitivity than RPS-BLAST. ### How current is the CDD database? CDD is updated periodically as new domain models are curated and imported from partner databases. The web tool always searches against the current release. If you run rpsblast locally, check the NCBI CDD FTP site for database updates and re-download periodically to ensure your local version matches the current release. ### How do I cite CDD in my publication? Cite the CDD database itself along with the specific tool you used. The primary CDD citations are available on the CDD homepage at `https://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml`. For the web CD-Search tool, cite the relevant NCBI Structure group publication listed there. If you used rpsblast locally, cite the BLAST+ publication as well. ## Sources 1. [Database resources of the National Center for Biotechnology Information.](https://pubmed.ncbi.nlm.nih.gov/19910364/) 2. [Epitope mapping of gp350/220 conserved domain of epstein barr virus to develop nasopharyngeal carcinoma (npc) vaccine.](https://pubmed.ncbi.nlm.nih.gov/22715303/) 3. [Targeting conserved N-glycosylation blocks SARS-CoV-2 variant infection in vitro.](https://pubmed.ncbi.nlm.nih.gov/34839261/) 4. [SARS-CoV 2 spike protein S1 subunit as an ideal target for stable vaccines: A bioinformatic study.](https://pubmed.ncbi.nlm.nih.gov/34307057/) 5. [Computational analysis of envelope glycoproteins from diverse geographical isolates of bovine leukemia virus identifies highly conserved peptide motifs.](https://pubmed.ncbi.nlm.nih.gov/29310678/) 6. [CRESS domain-based molecular characterization of Begomovirus gossypimultanense.](https://pubmed.ncbi.nlm.nih.gov/42461357/) 7. [Evolution of the Large Nucleocytoplasmic DNA Viruses of Eukaryotes and Convergent Origins of Viral Gigantism.](https://pubmed.ncbi.nlm.nih.gov/30635076/) 8. [Origin of giant viruses from smaller DNA viruses not from a fourth domain of cellular life.](https://pubmed.ncbi.nlm.nih.gov/25042053/) 9. [Chordopoxvirus protein F12 implicated in enveloped virion morphogenesis is an inactivated DNA polymerase.](https://pubmed.ncbi.nlm.nih.gov/25374149/) 10. [Extensive domain shuffling in transcription regulators of DNA viruses and implications for the origin of fungal APSES transcription factors.](https://pubmed.ncbi.nlm.nih.gov/11897024/) 11. [COVID-19 Pathogen Viral Evolution Leading to Increased Infectivity.](https://pubmed.ncbi.nlm.nih.gov/35949765/) 12. [Comprehensive Evolutionary and Structural Analysis of the H5N1 Clade 2.4.3.4b Influenza a Virus Based on the Sequences and Data Mining of the Hemagglutinin, Nucleoprotein and Neuraminidase Genes Across Multiple Hosts.](https://pubmed.ncbi.nlm.nih.gov/41011764/) 13. [Identification of a retinoic acid-inducible gene I from Japanese eel (Anguilla japonica) and expression analysis in vivo and in vitro.](https://pubmed.ncbi.nlm.nih.gov/27238428/) 14. [Molecular analysis of capsid protein of Homalodisca coagulata Virus-1, a new leafhopper-infecting virus from the glassy-winged sharpshooter, Homalodisca coagulata.](https://pubmed.ncbi.nlm.nih.gov/19537993/) 15. [Structural Mapping of Surveillance Data Reveals Conservation of NNI Binding Site in RSV L Protein.](https://pubmed.ncbi.nlm.nih.gov/41599069/) 16. [Targeted immune epitope prediction to HHLA2 and MAGEB5 protein variants as therapeutic approach to related viral diseases.](https://pubmed.ncbi.nlm.nih.gov/34320928/) 17. [Computer-aided drug design against spike glycoprotein of SARS-CoV-2 to aid COVID-19 treatment.](https://pubmed.ncbi.nlm.nih.gov/33083627/) 18. [The SWIRM domain: a conserved module found in chromosomal proteins points to novel chromatin-modifying activities.](https://pubmed.ncbi.nlm.nih.gov/12186646/) 19. [The eukaryotic translation initiation regulator CDC123 defines a divergent clade of ATP-grasp enzymes with a predicted role in novel protein modifications.](https://pubmed.ncbi.nlm.nih.gov/25976611/) 20. [Integration of an anellovirus genome in the SKNO-1 acute myeloid leukemia cell line.](https://pubmed.ncbi.nlm.nih.gov/42671192/)