The problem with identical labels still exists.

Please resolve.

Users should be able to filter for specific datasets. We should use that to select (similar to gene select but for multiple) fewer datasets, which will lead to fewer rows by default and hence a faster startup.

Hi @boehmv,

I would like to comment and document on expanding the number of datasets. So far, I have processed data for phases 1 and 2 (all the cell types). These were systematically analyzed. The other datasets were not analyzed . I did try to figure out a way to quickly, for example, using a preprocessed server (such as recount3, among others), but this was impossible because the server misses most of the datasets.

What I have on my list:

• RNPS1 KD
• CASC3 KD
• UPF1 KD, SMG6, and SMG7 KD and double KD and long reads

Feel free to comment so we can update this. This is a "long-running" issue, meaning this will not be resolved in the short term, but we should aim to expand as much as we can.

The app is rendering URL for missing links for transcripts that are not in the DB GFF file, for example, MSTRG.13046.2.1.

Currently the database queries are not using indexes, as this is not directly supported by dbplyr:
Here is workaround tidyverse/dbplyr#484 (comment)

https://support.openanalytics.eu/t/create-running-user-proxies-containers-in-advance/2521/2

The table in the transcript view becomes a little bit hard to navigate if many transcripts and ORFs exist and many contrast are chosen. A heatmap or dotplot next to the transcript structures would be a nice enhancement.

gray dots = not significant
else: fold-change as color and -log(p-value) as dot size

What do you think ?

Working with Renv and Docker presents some challenges. Here are the key problems and solutions I am currently using:

• Incomplete Package Installation: Not all required packages are installed during renv::restore(). For instance, renv::install('markdown') needs manual execution, despite being listed in renv.lock.
• Snapshot Limitations: The command renv::snapshot(type = 'explicit') fails to include all packages specified in the Description file.
• Extended Cache Traversal Time: Occasionally, Renv takes over 30 minutes to traverse the cache directory, significantly slowing down the build process. Normally, Docker builds take about 5 minutes when packages are cached. A workaround involves creating a new renv.lock file using renv::snapshot(type = 'explicit') and manually installing the missing packages.

For later release stage, I would recommend understanding this problem and maybe replacing the docker base by a more adequate one.

I drafted a new logo and can generate a favicon from it:

The green blob is a Pac-Man and the red circle over the t is a lollipop that we use to represent the stop codons of a PTC.
Not full resolution / transparent yet, but if you have any feedback, I can change now.

• #25
• #26
• Tutorial is still missing
• Filter out missing CDS from the transcript table
• Filter genes without mRNA transcripts
• #24
• #23
• Dotplot and transcript structure are unnaligned for transcripts and some combinations of sources (see SRSF2)

Navigate to BAG1 locus.

Choose all ORF sources..

In this graphic it says no PTC but in the figure below it says PTC

Dynamically showing multiple contrasts has been a challenge. We need to find a simple way to show isoform usage + isoform structure.

• Christoph has suggested barcode plots.

Scrolling up and down to understand which ORF originates from where is not a great option

Hi @tbrittoborges

I noticed a problematic ORF for BAG3 based on EnsEMBL transcript ENST00000369085
which is living under the EnsEMBL annotation but does not match the EnsEMBL annotation

Replace the heatmap plot for genes with high number of transcripts by a download btn, as these take a long time to render and don't look good. This depends on #13.