Protein Knowledge and Research Tools: IDG/Pharos

Estimated time: 20 minutes

Use Pharos to compare protein annotations, Target Development Levels, publications, interactions, ligands, and drugs for the 25 genes.

Protein annotations

GTEx and HuBMAP showed where the genes have indexed expression. Pharos shows what is known about the encoded proteins and which research tools may exist.

IDG stands for Illuminating the Druggable Genome. This NIH Common Fund program develops knowledge and research tools for understudied proteins in druggable protein families. Its Pharos resource combines protein identifiers, annotations, publications, interactions, ligands, drugs, and target-development categories.

Why use Pharos for this paper?

The paper reports candidate genetic variants in 25 protein-coding genes. We use the following fields to compare existing knowledge and available research tools for each encoded protein:

Field Why it is included
Gene symbol and UniProt ID Connect the paper’s gene to a specific protein target.
Target Development Level Summarize the target’s Pharos development category.
Publication count Show the number of publications linked to the target.
Protein-interaction count Show the number of interaction records aggregated by Pharos.
Ligand and drug counts Identify targets whose chemical records warrant closer inspection.

Learn more about the available GraphQL fields in the Pharos API documentation.

Target development levels

Pharos uses four Target Development Levels, called TDLs:

TDL Pharos criteria How we use it here
Tclin An approved drug acts through the target. Open the drug records and examine their mechanisms and indications.
Tchem Among targets outside Tclin, the target has high-potency small-molecule binding data. Examine the ligand records for potency, selectivity, and experimental use.
Tbio The target has experimental Gene Ontology annotations or meets Pharos criteria based on publication, GeneRIF, and antibody records. Begin with mechanism, pathway, phenotype, or assay information.
Tdark Available knowledge is limited under the Pharos criteria. Begin by describing basic protein function and available reagents.

Scope of target development levels

TDL describes a protein. The study classification describes a variant.

TipWhy Pharos helps

The expression resources help us select a biological system. Pharos shows the protein knowledge, chemical tools, and clinical relationships already recorded for a target.

Querying the Pharos API

Pharos uses GraphQL. The helper requests selected target fields for each of the 25 gene symbols and combines the responses in one table.

Prepare the gene list and API helper

Load the published variants, select the 25 unique gene symbols, and import the Pharos wrapper.

gene_symbols contains the 25 unique gene symbols in alphabetical order. variants contains all 54 published rows.

Request protein annotations

Request protein-target records for the 25 genes and display the first five targets. In the returned table, compare TDL with the separate publication, interaction, ligand, and drug counts.

One row represents one protein target. The gene symbol and UniProt identifier link it to Pharos.

The other fields report its TDL and counts of publications, protein interactions, ligands, and drugs.

TDL follows Pharos development criteria. The counts describe records attached to the target. A Tchem target can therefore have ligand records when its drug count is zero.

NoteLive data

Target records are requested directly from Pharos. Target information and service availability can change.

How the wrapper works

The query requests only the protein fields used in this lesson:

query TargetContext($symbol: String!) {
  target(q: {sym: $symbol}) {
    sym
    name
    uniprot
    tdl
    publicationCount
    ligandCounts { name value }
    ppiCounts { name value }
  }
}

$symbol is a query variable. For MYLK3, the wrapper sends {"symbol": "MYLK3"} separately from the query text.

Pharos returns the selected fields under data and target. The wrapper extracts those fields and creates one table row per gene.

The complete implementation is in api_helpers.py.

Review the protein targets

Summarize the Target Development Levels, then inspect the publication, interaction, ligand, and drug records.

Summarize target development levels

Count the Target Development Levels in the current Pharos response. Complete the pandas method that counts each distinct category.

Use the pandas Series method that counts each distinct value.

# Count genes by TDL.
tdl_summary = (
    pharos["tdl"]
    .value_counts()
    .rename_axis("tdl")
    .reset_index(name="genes")
)
tdl_summary

The genes column counts how many queried targets fall into each Pharos TDL. The dated teaching data contain 18 Tbio, 4 Tclin, 3 Tchem, and zero Tdark targets.

The Tclin targets are TTR, DMD, KCNQ1, and MYH7. The Tchem targets are MYLK3, POLG, and TNNI3K.

Compare target fields

Keep the fields used for interpretation and order the proteins by TDL, drug count, and gene symbol. Include ligand counts so the table shows both ligand and drug records for Tchem targets.

TDL and linked record counts suggest different starting points:

  • TTR is Tclin with 41 ligand and 11 drug records. Open its Pharos target record to inspect the named drugs, mechanisms, and indications.
  • MYLK3 is Tchem with five ligand and one drug record. Its chemical records are the next place to review potency, selectivity, and possible experimental use.
  • TNNT2 is Tbio with 441 linked publications and 167 protein-interaction records. These records point to known protein biology and interactions as starting points for follow-up.
  • POLG is Tchem with two ligand records and zero drug records. Its qualifying chemical activity and its drug count are separate parts of the target record.

Find targets with drug records

Select Tclin and Tchem targets with at least one drug record.

The filter returns all four Tclin targets, TTR, DMD, KCNQ1, and MYH7, along with the Tchem targets MYLK3 and TNNI3K. The complete table records two ligand records and a drug count of zero for POLG.

Check your understanding

✔

Why can a Tchem target have ligand records but a drug count of zero?

Correct. Tchem reflects qualifying small-molecule activity, while the count fields summarize ligand and drug records attached to the target.

Ligand and drug records are counted separately in the Pharos response.

Tchem specifically identifies qualifying small-molecule activity.

✔

Which field describes protein target development alongside the paper’s classification of a specific variant?

Correct. TDL belongs to the protein target record. study_class records the paper’s classification of the variant.

study_class is the paper’s classification of a specific variant.

Key points

  • The dated Pharos response contains 18 Tbio, 4 Tclin, and 3 Tchem targets.
  • TDL, publication counts, interaction counts, ligand counts, and drug counts describe different parts of a protein target record.
  • Six Tclin or Tchem targets have at least one drug record in the dated response; POLG is Tchem with ligand records and a drug count of zero.

Next: Join the GTEx, HuBMAP, and Pharos results to all 54 variant rows and prioritize research follow-up.

View Jupyter notebook