Daily Bits - March 2023 | Sam’s Notebook

20230329

Completely nuked raven, but managed to recover - with improvements!. Here’s how it happened:
1. Couldn’t run R Markdown code that I had tested on my laptop on RStudio Server on raven. Required updating of tidyverse, but the update always failed when trying to install a dependency (ragg).
2. Tried to resolve ragg installation failure led to me down a rabbit hole of trying to figure out why jpeglib.h kept triggering an error. Attmempted to install/re-install numerous Linux jpeg libraries to no avail.
- During this process, I came across this command apt policy ubuntu-desktop which revealed that raven didn’t have a “desktop” installed! Could this explain why raven doesn’t boot to the GUI login screen??!!
  - Tried this as a potential solution: sudo apt-get install --reinstall ubuntu-desktop. When I ran apt policy ubuntu-desktop after that, the output now shows a desktop is installed!! Can’t wait to swing by lab to see if it worked!
1. Finally figured I’d remove and uninstall the system library associated with jpgelib.h: sudo apt remove libjpeg-dev. The amount of things that scrolled through the screen, and the time it took (~10 minutes), made it seem like it was removing many/most of the system programs (e.g. Thunderbird, Gnome stuff, …). Not good!
2. Found this post on StackOverflow, which ironically was where I got the idea to try to remove that libjpeg-dev library, and ran sudo apt install ubuntu-desktop. This also took a very long time (15 minutes), but became clear it was reinstalling all of the stuff that had just been removed. Whew!
3. After that, went to use RStudio Server and could no longer access. Gah!
4. After a bunch of rigamarole, poking around RStudio Server config files and whatnot, I just decided to follow the instructions (which are clearly writen and explained) on the Posit (RStudio) website.
5. Unfortunately, that didn’t prove to be the full crux of the problem. Now, even though RStudio Server would start, it couldn’t find the R installation (even though it was showing the correct path from the previous R installation). It turns out, the previous R installation didn’t survive that sudo apt remove libjpeg-dev command…
6. Followed instructions for installing R binaries on Ubuntu and specifiying the install location. Updated /etc/rstudio/rstudio.conf with the new R version path, and boom! Back in business!!
7. Bonus: Updated versions of R and RStudio Server. The later now enables use of Quarto within RStudio Server.
Finally, finally managed to finish updating “cannonical” P.generosa anotation file with gene ontology (GO)aspect info!

20230328

Worked on updating “cannonical” P.generosa anotation file with gene ontology (GO)aspect info by creating a singlur R Project/repo, instead of bouncing between a Jupyter Notebook and R. Additionally, decided to abondon trying to work within the GSEAbase package to map gene IDs with GO/GOslim IDs. Resulted in taking the “old fashioned” approach of creating a large “flat file” for joining/merging.

20230322

Olivia’s NSA practice talk.
Made progress on this GitHub Issue asking me to update our “cannonical” P.generosa anotation file with gene ontology (GO)aspect info.
- Remembered that I don’t actually need to use the updated UniProt programmatic API (which I’ve had problems getting to work), since I saved the output https://gannet.fish.washington.edu/Atumefaciens/20220419-pgen-gene_annotation_mapping/20220419-pgen-uniprot_batch-results.txt when I originally generated the original annotaiton file.
- Figured out how to parse/organize gene ontology (GO) aspect (biological process, cellular component, molecular function) from the previous UniProt batch retrieval file.
Continued to help Danielle Becker (Putnam Lab) with her WGBS analysis using nf-core/methylseq pipeline.
- This time (using the WGBS data instead of RNA-seq) I noticed that all of her R2 sequences (and none of her R1 seqs) showed overrepresented sequences in the MultiQC report. I suspected this might by polyG sequences (a known issue with Illumina NextSeq) and reviewed her FastQC for the R2 reads. Turns out, I was right. So, suggested she re-trim and enable polyG detection in whichever trimming software she’s using.

20230321

Helped out Danielle Becker (Putnam Lab) figure out issue(s) with her nf-core/methylseq pipeline. She is/was testing alignment of P.meandrina WGBS reads to P.verrucosa (which used to be the closest geneome available) and P.meandrina (which recently became available).
- Her issue was the she was using her RNA-seq data instead of her WGBS data…
Still can’t figure out the UniProt programmatic retrieval issue.

20230320

Read Ch.1 & 2 of “Invisible Women” for lab meeting.
Lab meeting.
Submitted for coral miRNA sequencing to genohub.
Still trying to figure out UniProt API. Responses from UniProt help desk were mostly unhelpful at this time.
CEABIGR meeting.
- Yaamini showed a bunch of dot plots of transcriptional noise.

20230318

Stopped in at lab and enabled remote desktop access on swoose.

20230317

-Science hour

Trying to figure out new UniProt API. Old Perl script to access UniProt reuturns a 405 error due to the fact that they’ve overhauled the API. Grrrr…. Getting scrambled output from query. Submitted help ticket to UniProt.

20230316

Messed around extensively trying to set up/use remote desktop access to raven and swoose. Was unsuccessful on both fronts. Need to enable remote desktop in person on swoose (although, I swear I’ve used remmina to do this in the past…). X11 forwarding is insufficient for the long run times of MEGAN6.
Pub-a-thon

20230315

Helped with Arianna downloading SRA data using SRA Tool Kit (GitHub Issue).
Discovered “fix” to inability to ssh into Mox after connection is broken. Need to exit the ssh master session:
```
  ssh -O exit mox
```

20230314

Spent most of the day attempting to gain remote desktop access and/or X11 forwarding on raven and swoose. Couldn’t get either to work on raven, as it’s likely tied to the inability to get raven to boot to the Ubuntu login screen (suggests a disply problem). On swoose X11 forwarding works (needed for MEGAN6), but remote desktop does not. This is an issue because it is taking hours to import DIAMOND BLASTx MEGANIZED-DAA files. If ssh connection is lost, then MEGAN6 is killed (even when using tmux).
Helped Steven and Matt with a weird problem Matt was having trying to get a PO number using the TREQ purchasing system. He never received a PO, but I was able to search Ariba and find that a PO had actually been issued… Provided him with PO.

20230313

Helped Javi get access to UW Husky OnNet for Raven usage for E5.
Added administrative instructions to the CEABIGR wiki to aid in setting up UW Husky OnNet access/provisioning.
Lab meeting.
CEABIGR meeting.

20230310

E5 bi-monthly meeting:
- Assigned to work on this GitHub Issue.
Worked on E5 deep dive developing workflow for isoform expression (GitHub Issue).
Science Hour.

20230309

Pub-a-thon
Continued working on this GitHub Issue to update our geoduck gene annotations file.

20230308

Spent all day (seriously) working on this GitHub Issue to update our geoduck gene annotations file to include GOslim and “aspect” (i.e. Biological Process, Molecular Function, or Celullar Component). Unfortunately, this is kind of complicated. Here’s the process I’m (attempting) to employ:
1. Get GOslims:
- Doing this using GSEAbase (R/Bioconductor package). Process is easy, but I feel there’s small bug where a GO ID doesn’t get mapped to “itself” if it’s a GOslim/parent ID.
- Keeping gene IDs so that they can remain associated with the input GO IDs. This was solved in this Bioconductor support forum.
1. Join with the existing annotation file (https://gannet.fish.washington.edu/Atumefaciens/20220419-pgen-gene_annotation_mapping/20220419-pgen-gene-accessions-gene_id-gene_name-gene_description-alt_gene_description-go_ids.tab).
- This should be relatively straight forward (hopefully I didn’t jinx myself!) once I’m able to get the “missing” GO/GOslim mapping problem resolved.

20230303

Science Hour
- Steven tested HiSat2 alignment rates when specifying (or not) exons and splice sites using some coral FastQs and the M.capitata genome. Result - very little difference (< 1%).
Oyster Epigenetics meeting w/Lotterhos Lab.

20230302

In lab
- Fixed an issue on Raven where an error message pops up during reboot indicating there’s a fan response failure. This was a problem because there was no way to bypass/acknowledge this error remotely, thus preventing reboot without someone being physically at the computer to acknowledge the error. Turns out, Dell has a support page on how to handle this issue, and the solution was to go into the Bios and turn off the HDD fans for fan power pins in which no fan was actually connected!
- Attempted to fix issue on Raven where the desktop is not accessible. After reboot, the screen is just a series of lines of boot info with an [OK] printed at the beginning of each line. This does not affect remote access via ssh, but prevents users from trying to use the computer when physically sitting at Raven. I suspect, but haven’t confirmed, that this also prevents remote desktop connections. Tried many, many things (altering /etc/default/grub, switching to lightdm from dgm3 and back, installing the correct video drivers for the video card - AMD Radeon PRO WX 2100, trying to boot in safe graphics mode) all to no avail.
  
  I did find a Dell knowledge base solution addressing this and the solution is to reinstall the OEM operating system with the monitor plugged into a different video card!

20230301

Transferred trimmed C.magister RNA-seq data from Google Drive to our HPC (Mox) using rclone. This was a great suggestion from Giles Goetz! Additionally, the rclone website has a great explanation on how to configure a connection to Google Drive. The key to doing this, though, was specifying --drive-shared-with-me option. Without that, the shared drive was not visible/accessible. In the code below, noaa-crab is the name of the configuration I set up following the instructions linked above.
```
  rclone-v1.61.1-linux-amd64/rclone \
  copy \
  --progress \
  --drive-shared-with-me \
  noaa-crab:202301-dungeness_crab-transcriptome/ \
  ./
```
Worked on, and finished, generating the comprehensive CEABIGR CSV Steven requested in this GitHub Issue.