Mustang Analytics
Computational target triage · ALS

46 ALS genes, one live evidence graph

This pipeline takes the 46-gene amyotrophic lateral sclerosis panel and pulls the evidence around the panel genes from six public databases: protein records from UniProt, interaction partners from STRING, classified variants from ClinVar, pathway membership from Reactome, literature from PubMed, and disease associations and known drugs from Open Targets. It links the records into one graph, scores the genes, and writes 589 candidate mechanisms tied to real citations. Rows trace back to cached API responses.

Built 2026-06-21 · Six public data sources · Reproducible · Code on GitHub

46Seed genesreal UniProt accessions
589Machine hypotheses244 distinct citations
2,424ClinVar variantsacross the 46 genes
110Open Targets drugsreal ChEMBL ids
What this is

An evidence graph drawn from public databases

The 46 seed genes resolve to UniProt accessions, then query across five more sources. The responses are cached on disk and normalized into a small database: genes, interactions, variants, pathways, disease associations, drugs, and papers. The graph that results carries 405 interaction edges, 2,424 variants, 163 Reactome pathways, and 362 papers with 244 distinct PubMed ids.

Wiping the cache and running again rebuilds the database to the same shape, which is the test that no row was seeded by hand. The full provenance, with spot-checked records and their source URLs, ships with the project.

Data sources

SourceRecordsEndpointWhat it provides
UniProt46 proteinsrest.uniprot.orggene to accession, name
STRING405 edgesstring-db.orginteractions, confidence ≥ 0.7
NCBI ClinVar2,424 variantseutils.ncbi.nlm.nih.govclinical classification
NCBI PubMed362 paperseutils.ncbi.nlm.nih.govliterature, citations
Reactome163 pathwaysreactome.orgpathway membership
Open Targets92 ALS links, 110 drugsplatform.opentargets.orgdisease association, known drugs
Ranked genes

The panel, scored on combined evidence

Genes are scored on a weighted blend of Open Targets association, ClinVar pathogenicity, pathway and network centrality, literature volume and citation quality, and druggability. The ALS association column is the Open Targets score for the gene against amyotrophic lateral sclerosis. Click a row for a plain-language breakdown, including the published share of familial and sporadic ALS and the phenotype specificity (from GeneReviews where available; formal per-gene sensitivity and specificity are not standard).

#GeneUniProtProteinALS assoc.Score
1TUBA4AP68366Tubulin alpha-4A chain0.7770.725
2ERBB4Q15303Receptor tyrosine-protein kinase erbB-40.6620.627
3SOD1P00441Superoxide dismutase [Cu-Zn]0.8700.605
4TBK1Q9UHD2Serine/threonine-protein kinase TBK10.6350.569
5VCPP55072Transitional endoplasmic reticulum ATPase0.6640.555
6SIGMAR1Q99720Sigma non-opioid intracellular receptor 10.7050.554
7SQSTM1Q13501Sequestosome-10.8090.533
8OPTNQ96CV9Optineurin0.7710.514
9PFN1P07737Profilin-10.5770.502
10FUSP35637RNA-binding protein FUS0.8120.497
11HNRNPA1P09651Heterogeneous nuclear ribonucleoprotein A10.6410.495
12ANGP03950Angiogenin0.7870.495
13CYLDQ9NQC7Ubiquitin carboxyl-terminal hydrolase CYLD0.0000.488
14UBQLN2Q9UHD9Ubiquilin-20.6090.487
15FIG4Q92562Polyphosphoinositide phosphatase0.5370.486
16KIF5AQ12840Kinesin heavy chain isoform 5A0.7080.482
17CCNFP41002Cyclin-F0.4180.482
18SPG11Q96JI7Spatacsin0.1980.477
19ANXA11P50995Annexin A110.7740.467
20TARDBPQ13148TAR DNA-binding protein 430.8200.464
21NEK1Q96PY6Serine/threonine-protein kinase Nek10.5810.464
22ALS2Q96Q42Alsin0.5850.459
23C9orf72Q96LT7Guanine nucleotide exchange factor C9orf720.6100.444
24SPTLC1O15269Serine palmitoyltransferase 10.3180.438
25CHMP2BQ9UQN3Charged multivesicular body protein 2b0.7730.427
26DCTN1Q14203Dynactin subunit 10.7290.416
27PRPHP41219Peripherin0.4670.408
28SETXQ7Z333Helicase senataxin0.3780.384
29ATXN2Q99700Ataxin-20.4440.354
30UNC13AQ9UPW8Protein unc-13 homolog A0.4690.330
31UBQLN4Q9NRR5Ubiquilin-40.4630.307
32VAPBO95292Vesicle-associated membrane protein-associated protein B/C0.4020.300
33TIA1P31483Cytotoxic granule associated RNA binding protein TIA10.0780.292
34MATR3P43243Matrin-30.7470.289
35LRP12Q9Y561Low-density lipoprotein receptor-related protein 120.3770.287
36CFAP410O43822Cilia- and flagella-associated protein 4100.5680.271
37MOBPQ13875Myelin-associated oligodendrocyte basic protein0.4440.251
38DAOP14920D-amino-acid oxidase0.4720.250
39FGGYQ96C11FGGY carbohydrate kinase domain-containing protein0.3720.238
40SCFD1Q8WVM8Sec1 family domain-containing protein 10.2850.236
41TMEM106BQ9NUM4Transmembrane protein 106B0.1170.227
42NEFHP12036Neurofilament heavy polypeptide0.4920.226
43TRPM7Q96QT4Transient receptor potential cation channel subfamily M member 70.0210.224
44CHCHD10Q8WYQ3Coiled-coil-helix-coiled-coil-helix domain-containing protein 10, mitochondrial0.4090.204
45ELP3Q9H9T3Elongator complex protein 30.2160.195
46TAF15Q92804TATA-binding protein-associated factor 2N0.0360.157

ALS assoc. is the Open Targets disease-association score (0 to 1). Score is the pipeline's combined gene rank.

Mapped drugs

Known drugs against the panel targets

Open Targets returns 110 drugs and clinical candidates whose molecular target is one of the panel genes, with real ChEMBL ids and development stages. TOFERSEN, the approved SOD1 antisense therapy, is among them. Most-advanced shown first. Click a row for the drug detail.

DrugChEMBLStageTargetMechanism
AfatinibCHEMBL1173655ApprovedERBB4Receptor protein-tyrosine kinase erbB-4 inhibitor
Afatinib DimaleateCHEMBL2105712ApprovedERBB4Receptor protein-tyrosine kinase erbB-4 inhibitor
Belantamab MafodotinCHEMBL4298209ApprovedTUBA4ATumor necrosis factor receptor superfamily member 17 binding agent
Brentuximab VedotinCHEMBL1742994ApprovedTUBA4ATubulin inhibitor
CabazitaxelCHEMBL1201748ApprovedTUBA4ATubulin inhibitor
CarbetapentaneCHEMBL73234ApprovedSIGMAR1Sigma opioid receptor agonist
Carbetapentane CitrateCHEMBL1256696ApprovedSIGMAR1Sigma opioid receptor agonist
ColchicineCHEMBL107ApprovedTUBA4ATubulin inhibitor
DacomitinibCHEMBL2105719ApprovedERBB4Epidermal growth factor receptor erbB1 inhibitor
DextromethorphanCHEMBL52440ApprovedSIGMAR1Sigma opioid receptor agonist
Dextromethorphan HydrobromideCHEMBL1256818ApprovedSIGMAR1Sigma opioid receptor agonist
Dextromethorphan PolistirexCHEMBL1201461ApprovedSIGMAR1Sigma opioid receptor agonist
DocetaxelCHEMBL3545252ApprovedTUBA4ATubulin inhibitor
Docetaxel AnhydrousCHEMBL92ApprovedTUBA4ATubulin stabiliser
Enfortumab VedotinCHEMBL3301589ApprovedTUBA4ANectin-4 binding agent
EribulinCHEMBL1683590ApprovedTUBA4ATubulin inhibitor
Eribulin MesylateCHEMBL1683544ApprovedTUBA4ATubulin inhibitor
FenfluramineCHEMBL87493ApprovedSIGMAR1Serotonin 2c (5-HT2c) receptor agonist

Showing 18 of 110 mapped drugs.

Shared pathways

Reactome pathways shared across panel genes

Click a row to see which panel genes share the pathway.

PathwayNamePanel genes
R-HSA-2132295MHC class II antigen presentation3
R-HSA-8951664Neddylation3
R-HSA-5205685PINK1-PRKN Mediated Mitophagy3
R-HSA-114608Platelet degranulation3
R-HSA-2565942Regulation of PLK1 Activity at G2/M Transition3
R-HSA-5357905Regulation of TNFR1 signaling3
R-HSA-5357786TNFR1-induced proapoptotic signaling3
R-HSA-8854518AURKA Activation by TPX22
R-HSA-936964Activation of IRF3, IRF7 mediated by TBK1, IKKε (IKBKE)2
R-HSA-9646399Aggrephagy2
R-HSA-5620912Anchoring of the basal body to the plasma membrane2
R-HSA-6811434COPI-dependent Golgi-to-ER retrograde traffic2
R-HSA-6811436COPI-independent Golgi-to-ER retrograde traffic2
R-HSA-6807878COPI-mediated anterograde transport2

163 distinct Reactome pathways total; the ones touched by the most panel genes are shown.

Generated hypotheses

The 589 candidate mechanisms, searchable

The pipeline writes one hypothesis per pattern it finds in the graph: two ALS genes sharing a cell process, two proteins that work together, or a gene with disease-causing variants that links to a known ALS gene. Many read alike, so open one to see what sets it apart and to walk through how the pipeline got there, step by step. Search the text, or filter by mechanism type and confidence.

Plain-language key
ALS
A disease that kills the nerve cells that control muscles. Also called Lou Gehrig's disease.
Gene
A stretch of DNA that holds the recipe for one protein.
Protein
The working part the recipe builds.
Pathway
A group of proteins that team up to finish one job in the cell. Source: Reactome.
Interaction
Two proteins that touch or affect one another. STRING scores how sure we are, from 0 to 1.
Variant
A spelling change in the DNA. ClinVar records whether the change is known to cause disease.
Association score
How tightly a gene is tied to a disease, from 0 to 1. Source: Open Targets.
PMID
The id of a paper in PubMed, the public library of medical research.
The panel as a network

How the 46 genes connect

Nodes are panel genes; lines are STRING interactions between two panel genes at confidence 0.7 or higher. A node grows with its betweenness, a measure of how often it sits on the shortest path between two other panel genes, so the bridge genes read largest. Rust marks a gene with an Open Targets ALS association of 0.5 or higher; pacific marks the rest. Pick a layout to redraw the network, drag a node to move it, or rotate the whole view. Hover a node to trace its links, or click one to open its row in the gene table above.

Layout
ALS2ANGANXA11ATXN2C9orf72CCNFCFAP410CHCHD10CHMP2BCYLDDAODCTN1ELP3ERBB4FGGYFIG4FUSHNRNPA1KIF5ALRP12MATR3MOBPNEFHNEK1OPTNPFN1PRPHSCFD1SETXSIGMAR1SOD1SPG11SPTLC1SQSTM1TAF15TARDBPTBK1TIA1TMEM106BTRPM7TUBA4AUBQLN2UBQLN4UNC13AVAPBVCP

37 STRING interactions run between the 46 panel genes. The full graph, with partners, variants, pathways, papers, and hypotheses, downloads as als_knowledge_graph.graphml for Gephi or Cytoscape.

Notes & provenance

Data comes from UniProt, STRING, NCBI ClinVar and PubMed, Reactome, and Open Targets. The public APIs are queried live and cached for reproducibility. Variant classifications use ClinVar germline classifications; drug stages come from Open Targets. This page is a worked example of an evidence-integration pipeline. It is informational and is not medical advice.