Documentation Getting Started Navigating Gene-Centered Results
Browse Documentation
Introduction to PlantQTLdb Searching for Genes and Traits Navigating Gene-Centered Results Using Trait–Gene Evidence Gene Prioritization Connect ChatGPT, Codex and Scripts

Getting Started

Navigating Gene-Centered Results

Move from the gene overview to molecular phenotypes, QTL evidence, and supporting literature.

Overview

The Overview summarizes gene annotation and the evidence types currently indexed for the selected gene. Molecular phenotype, molecular QTL, nearby GWAS, and literature counts use different units and should be read independently.

Supporting result layers

Molecular Phenotypes
Expression, alternative splicing, and alternative polyadenylation across indexed datasets.
QTL Evidence
Local molecular QTL and GWAS signals around the gene, plus full association explorers.
Literature Evidence
Read indexed gene–publication relationships and additional online candidates. Open the article, abstract and original evidence excerpts to check what each source supports.

Read the molecular measurements

Violin plots summarize available sample values. The sample table exposes individual measurements and source metadata. Column headings support sorting; filter controls provide categorical choices, text matching or numeric ranges as appropriate. Download the filtered rows when you need a reproducible sample selection. Population and subgroup labels follow each study; rice-specific terms such as indica and japonica are not assigned to other species.

Expression · FPKM
Gene expression in fragments per kilobase of transcript per million mapped reads. These displayed values are not the rank-normalized phenotypes used for QTL mapping.
Splicing · PSI
Estimated usage of the selected junction or retained-intron component, on a 0–1 scale. Select the dataset and event, and compare the same component. Different junction definitions are not interchangeable measurements.
APA · PDUI
Estimated distal poly(A)-site usage, on a 0–1 scale: higher values indicate greater relative use of the long 3′ UTR isoform in the fitted two-isoform model. DaPars2 infers the proximal site; the distal reference boundary comes from the analysis region.
APA region and site
A shared analysis interval can have different inferred proximal sites in different datasets. A grouped region is not proof of an identical cleavage-site pair. Read the dataset-specific site coordinates alongside PDUI; the interval alone is not a directly measured isoform.

Inspect read coverage

  1. Select samplesIn the sample table, choose up to eight samples with registered read tracks, then select View read coverage. Expression uses one gene, APA uses one region, and splicing uses one event. Read tracks are available only for connected datasets and samples.
  2. Choose quantile examplesSelect quantile examples chooses up to four eligible samples near the 10th, 25th, 75th and 90th percentiles within the current table filters, including multiple datasets when eligible. For splicing, choose one junction first. These are illustrative values, not population-representative samples or a differential-analysis result. Use Dataset filters for comparisons within a study.
  3. Adjust the viewWindow selects the full gene or selected event/region; Flanks adds 0, 100, 500 or 1,000 bp on each side. Use the mouse wheel to zoom, drag horizontally to pan, and double-click a position to zoom in. Reset view restores the chosen window. Download SVG saves the displayed figure.
Coverage and junctions
Read depth is mean aligned-base depth per displayed bin. Per million mapped alignments divides coverage by the mapped-alignment count in millions. Junction arcs connect flanking exon bases; their labels are supporting read counts. Minimum junction reads filters arcs, and hovering reveals counts hidden to prevent label overlap.
Highlight and annotation
The shaded region marks the selected AS event, and highlighted arcs are its junctions; other observed junctions provide context. Transcript tracks distinguish CDS, 5′ UTR and 3′ UTR. APA markers distinguish the predicted proximal site from the distal reference boundary.
Which reads contribute to the plot?

Primary alignments with MAPQ ≥ 20 are retained; duplicate and QC-failed reads are excluded. A junction needs at least eight aligned bases on both sides. Mates are counted separately and both strands are combined. Where a phenotype combines libraries, their alignment evidence is pooled. Library-depth scaling uses the total mapped-alignment count, including alignments excluded by the display filters.

Why do junction counts differ from PSI or PDUI?

Junction labels count supporting alignments. PSI and PDUI are imported model estimates, not values recalculated from the displayed arcs. Coverage helps inspect RNA-seq support, but does not by itself verify an isoform or exact cleavage site.

Can I compare samples from different datasets?

The viewer supports eligible samples across datasets. Match the species, reference and molecular feature, then consider tissue, treatment and library preparation. Library-depth scaling does not remove batch effects, and quantile examples do not establish a biological group difference.

Interpret literature counts and retrieval

Indexed publications can appear before missing external metadata are available. Titles, abstracts or dates may be filled in later; a temporarily incomplete record does not mean the publication is absent. Source availability can limit online retrieval. Publication dates prefer the electronic date when available and retain the precision reported by the source.

KG publications
Unique publications linked to this gene in the knowledge graph. One publication can support multiple indexed relationships; relationship and publication counts therefore differ.
Online candidates loaded
Unique candidates retrieved so far from Europe PMC, PubMed, PMC and OpenAlex. The parenthesized count covers candidates linked to the current gene in the KG, not all KG publications. Use Load more results to continue beyond the initial batch.
Relevance and sources
Title/abstract matches, full-text mentions, reviews and unverified matches describe retrieval evidence. KG membership, linkage to this gene and experimental support are separate checks. Read the original article before treating a search hit as biological evidence.
How should I interpret an extracted literature relationship?

The published extraction prompt requires a supporting text excerpt, species-aware gene identity and one gene per subject. It allows both positive and no-effect findings. Check the original text, experimental context and direction of the finding; an indexed relationship is not automatically an experimentally validated causal effect. The prompt documents the extraction rules, not proof that every extracted record is correct.

When a result is missing

Why is a gene not found?

Confirm the selected species and reference in the Gene Catalog. Direct search accepts supported identifiers and exact gene-name tokens; it does not search free-text annotation descriptions. A transcript ID or an ID from a different assembly may not match.

Does an empty evidence panel mean there is no biological effect?

First check the dataset and active filters. No significant result, missing input, unavailable analysis and QC excluded are different states. An empty panel alone cannot distinguish a true absence of effect from limited data or power.

Why is read coverage unavailable for a sample?

A molecular measurement can exist without a connected alignment file. Coverage requires a registered read track for that sample and a compatible reference. Try another eligible sample; absence of a track does not invalidate its recorded FPKM, PSI or PDUI.