wiki:xQTLUserReview

Version 6 (modified by jvelde, 13 years ago) (diff)

--

Report builder feedback 18 nov 2011

Yang:

  • Use but does not help you to 'get to know' the data in the first place.
  • Wanted:
    • Quantative data: boxplot of measurements for each sample/individual, to inspect and detect/remove outliers -> Use e.g. boxplot(log(metaboliteexpression, exp(2.71828))) -> Or boxplot(log(t(metaboliteexpression), exp(2.71828))) for the (transposed) 'other view'
    • Qualitative data: heatmap like view of e.g. genotypes -> Use e.g. image(matrix(as.numeric(as.factor(genotypes)),nrow(genotypes),ncol(genotypes))) -> Or heatmap(matrix(as.numeric(as.factor(genotypes)),nrow(genotypes),ncol(genotypes)),scale="none",Colv=NA, Rowv=NA)

Basten:

  • Enter gene -> get all eQTL information! Simple as can be.
  • Cis/trans information
  • Find genes in a region and get their QTLS (sounds like gBrowse)
  • Ofcourse everything in document 'concept map'

Matthias:

  • Need a bit more than just 'gene -> QTL' info, so like the Advanced button in the simple screen
  • Select a matrix with QTL results, then grab a row or column (e.g. select a trait)
  • Filter within that list using e.g. a threshold to get a smaller list of e.g. significant markers
  • Using this trait and marker, build a nice table with a report
  • Even more advanced: cross-datamatrix filtering. Select an individual and get from all matrices values which adhere to a certain filter.
  • Relation between promotor-gene-transcript-protein-enzyme-pathways should be used preferably to create a biological valuable report
  • gBrowse as an exploration tool would be super to have, and is reusable for many organisms

Panacea xQTL review Wed 9 nov 2011

  • Data curation and annotation of the Panacea Database (joint effort)
  • Host project analyses:
    • Phase 1: Just add the R scripts as files for basic provenance
    • Phase 2: Being able to rerun important scripts such as sample mislabeling and QTL x Env
    • Phase 3: Make it easy to add and run any R script that was used
  • Data matrix:
    • Download or visualize with CytoScape
    • Need new view with more organization/hierarchy: group by experiment or annotation
    • Need to couple this hierarchy with the ability to quickly run scripts for visualization or statistics on a piece of the data
  • Need 'supersearch' which produces reports on concepts (ie. everything related to a marker or individual)
    • This should include a special 'QTL finder' tool to quickly create reports on findings using a matrix and/or traits as inputs
  • Focus on pathways:
    • Want to query pathway information for an organism from 1..N sources (GO, WormBase, KEGG.. ?)
    • Want to couple existing gene annotations to these pathways, semi-automatically (?)
    • Want to use this information to run analysis/visualizations on these batches of genes
    • Want to create pathway plots (biology!) with the gathered information (e.g. QTL profiles)
    • Want to create CytoScape graphs of e.g. correlation of the genes

EURATRANS xQTL review Wed 9 nov 2011

  • Integrate with genome browser (gBrowse, formats WGL and GFF)
  • Need to save any filter as a URL
  • Need to clarify how a tool like gBrowse can specify a 'range' in filter URL syntax and link it
  • Also need the 'range' to work on matrices with locus data (e.g. markers) via URL
  • Need to add scripts as visualizers of data: e.g. let them appear in a list of possible plot types in matrix
  • Search box bugs: no '_name' searches possible
  • Use real column names instead of labels eventually, this is still very confusing
  • Need for advanced matrix views:
    • On similar rows & columns, create 'subheaders' with multivalue display
    • Alternatively, matrix merge operation to concatenate them
    • Want to have a report on a value that appears in multiple matrices
    • And/or reports with plots for e.g. a marker or probe with all related information from all sources

xQTL LL user workshop Fri 28 Oct 2011 general comments

  • Need more download formats: STATA, SAS, CSV (not TSV)
  • SNP data should somehow include metadata such as: imputed (yes/no), genome build, reliability, dosage, etc.
  • There should be a way to merge (part of a / a filtered) phenotype set with a genotype set, and retrieve the result
  • The application can be slow, need more speed
  • There are too many tabs, GUI is unclear
  • Need an hourglass (or something) to indicate the application is busy (blackout screen with progress bar when busy, GeneNetwork style?)
  • Making selections is too complex
  • Column paging is confusing
  • Filtering works 'half' (Joris please explain)
  • Data viewer should be able to create selections using your own file
  • Id's should be unrecognizable
  • Search filter should work on all measurements by default
  • Filters are lost when a user pages, this is annoying
  • Like other waiting times, applying a filter should cause the app to refresh with a 'wait please' status
  • Merging pheno- and geno sets is a special case of using a shopping cart (batches) to save selections for re-use elsewhere, this should be implemented. So you make a selection of participants based on phenotypic criteria, and then you view SNPs of interest for only those selected, and vice versa.