2022 Functional Genomics Lab Notebook

2022

2022-01 January

2022-01-21 Friday

[2022-01-21 Fri 11:39] Setting up new server

So some issues. There's no power cable for the ssd. And I don't have internet access on the server yet. There's also no deb for megacli to download so I had to use alien to create it. But alien isn't packaged up for nixos, so I had to make a docker image and run alien in there, then put that on a usb and then move it to the server.

Basic overview: https://gist.github.com/fxkraus/595ab82e07cd6f8e057d31bc0bc5e779

Thought I was going to have to order a power cable for the loose ssd, but I found one on the power supply.

2022-02 February

2022-02-19 Saturday Goals and Planning

Accomplishments

2021

• nextflow groseq rewrite and extension

• First biotech internship

• Got a pre-print published

• Mentored 3 undergrads

• Helped write and teach first course

TODO Previous 5 years

• Passed core courses

• Got heavily involved in nf-core

Research Goals

2022

• Finish QE

• Publish pipeline paper

• Finish infection GRO-Seq paper

• Start on GRO-seq mining

• Scale Suhana's Protein analysis

• Help Alyssa with WGS viral analysis

TODO Next 3 years

• ?

• Figure out direction

• Get some data sequenced by Element(A rich dataset like the GROseq)

Professional & Personal Goals

2022

• Learn Julia

• Become an nf-core core member

• Iterate on Applied Genomics

  • Look at GitHub Classroom(to automate feedback)

  • Pick datasets

  • Rework second half to expand projects

Next 3 years

• Come up reproducible "data science" workflow(nextflow level ease of use)

  • Can an undergrad reproduce it in a week?

  • Can an undergrad pick it up in a semester?

  • More "flexible" than nextflow

• Graduate

• Explore industry jobs(I'm really loving the "Genomics Applications" team)

2022-02-21 Monday

[2022-02-21 Mon 13:15] Goals and Planning Meeting

• GRO-cap might be another type of dataset to look at.

• Question to answer is what happens to the cell?

Three parts to my project

• Resource Generation (analysis DB)

• Toolkit (nf-core pipeline)

• Dynamics of eRNAs

  • How do they shutoff?

  • Negative feedback loop, product knocks down the TF

Random thoughts

• Peng looked for things going up and INFB was going down

  • Only found chromatin genes

• Reinfection data showed that IMR had a "memory" and didn't have a huge viral response, where GM didn't and had the exact same viral response

• scATAC-seq would be a good dataset of this to have to mirror the timecourse

• HiC does a first pass of sequencing to check for quality at a low sequencing depth, and then sequences the libraries with good quality at high-depth

• The lack of exon junction causes the eRNAs to be degraded. The exon-junction is usually a QC check in the nucleus to make sure bad RNAs don't get exported. The eRNAs are high jacking that system to get broken down.

Applied Genomics

• 3 different "modules", that are 3 weeks each.

• Different Groups can run the pipelines with different params(aligners, reference genome, GTF) to demonstrate the pros and cons of different methods, reproducibility issues, and just how big of an effect it can have.

• Cover QC concepts, basic alignment, and gene counts

2022-02-26 Saturday

[2022-02-26 Sat 20:38] Setting up raid controller and Julia

Issue with not being able to use megacli, was that I wasn't running it with sudo

sudo ln -s /opt/MegaRAID/MegaCli/MegaCli64 /usr/local/bin/megacli
sudo megacli PDList -aALL

To use the drives connected to the 2108/2208 controller, a RAID must be created. If using drives like in JBOD, each single drive must be created as a RAID 0 individually

Once again confirmed, going to need to make each disk a RAID 0

Installed storcli

wget https://downloadmirror.intel.com/685225/StorCLI_007.1704.0000.0000.zip
unzip StorCLI_007.1704.0000.0000.zip
sudo dpkg -i StorCLI_007.1704.0000.0000/Ubuntu/storcli_007.1704.0000.0000_all.deb
sudo ln -s /opt/MegaRAID/storcli/storcli64 /usr/local/sbin/storcli

Used storcli to setup disks

MegaRAID StorCLI — ASERGO Knowledge Base

# sudo -i
storcli /c0 set jbod=on
# Controller does not support JBOD
# :(

storcli /call/dall show all

storcli /c0 add vd r0 drives=245:0

for i in {4..12}; do
    storcli /c0 add vd r0 drives=245:$i
done

storcli /call/vall show

ZFS pool creation

• Going for mirrored drives and then ssd caching

sudo zpool create tank mirror /dev/sdd /dev/sde

Not sure if they should be Cache devices or SLOGs?

Porque no los dos?

Going to have a mirrored SLOG(To prevent data lose) and just throw both of them at the L2ARC cache, since it's just for reference and not for actual data. Might use them as a second SLOG

sudo zpool add -f tank log mirror /dev/sdj /dev/sdk
sudo zpool add -f tank cache /dev/sdl /dev/sdm

Made a scratch pool

sudo zfs create tank/scratch
sudo zfs set mountpoint=/scratch tank/scratch
sudo chown -R emiller:users /scratch/

2022-03 March

2022-03-07 Monday

[2022-03-07 Mon 10:00] Figure out nature of Transcripts   meeting

With Tae Hoon Kim

• Talked about PINTS

• Build a model for nascent RNAs to pull out the important pieces of eRNAs

• We need to get the actual eRNA transcripts, the TREs they provide are just identified enhancers.

  • Finding the ends of the eRNAs (We know the TSS)

    • Two ways, the "pausing" at the end, or the Poly A tail of RNA-seq

What the workflow will look like

• Pull in TREs

• Find 5' and 3' from GRO-seq

• Intersect of BAM and TRE, so the TRE is going possibly be too long, and the eRNA stops before the end of the enhancer

Interesting eRNAs will be:

• Highly Transcribed

• Long

Useful LSI MegaCli64 Commands – Help Center

2022-03-20 Sunday

[2022-03-20 Sun 21:55] Ran groseq transcripts against TREs

There's a ton of reads still. Basically, there were still reads, so yay!

2022-04 April

2022-04-04 Monday

[2022-04-04 Mon 11:17] Meeting with Tae

Updates

• Possible collaboration as an nf-core pipeline for viral integration in human NGS data

• Playing around with TRE

• Planning on starting Applied Genomics next week

Notes

• Possible collaboration with his friend from Yale looking at Nascent data in cancers. Looking specifically at Hox genes(The Role of HOX Transcription Factors in Cancer Predisposition and Progressio...) expression and nascent transcripts around there for cancers that have lost their positional inactivity.

• Write about PINTS and why it's better

2022-04-10 Sunday

[2022-04-10 Sun 22:12] Funnel Plot

import plotly.express as px
import pandas as pd

stages = [
    "All unannotated transcripts",
    "Transcripts that overlap with H3K4me1 and H3K27ac histone modifications",
    "Transcripts that satisfy cutoff for amplitude index and continuity index",
]
df_mtl = pd.DataFrame(dict(number=[127249, 21966, 10695], stage=stages))
df_mtl["Cell"] = "GM"
df_toronto = pd.DataFrame(dict(number=[88540, 34543, 14775], stage=stages))
df_toronto["Cell"] = "IMR"
df = pd.concat([df_mtl, df_toronto], axis=0)
fig = px.funnel(df, x="number", y="stage", color="Cell")
fig.show()

2022-04-18 Monday

[2022-04-18 Mon 10:00] Meeting with Tae

Feedback on Current Research Presentations

• Good on the less technical bit on presentation

• How aligners might perform

  • I should look into other aligner benchmarking

  • Look into STAR aligner paper on using an individualized genome to improve performance

  • The issue is repeats

• Ancient vs recent repeats

  • Recent are more difficult because they haven't picked up their own unique small differences

• 150bp reads are an issue because there's too many repeats

• ML model to fill in the transcripts

  • There might be transcripts on either side of a repeat, and take those multiple mapped regions and see if there was one in the middle

• PINTS subworkflow/workflow of nascent?

• We take RNA-Seq best practices for granted

• RNA polymerase (RNAP) is intimently linked to regulation

• What happens to the RNAs that get spliced?

Applied Genomics

• 3 Encapsulated modules

• Hybrid?