Lightnews — Scholar-powered news

Chris Miller @chrismiller.science · 1d

PSA: dbGaP authorized access/run selectors seem to be not working. I assume it's the shutdown - either they turned the lights off, or something's busted and no one is there to fix it

a man in a suit and tie holds his head in his hands

ALT: a man in a suit and tie holds his head in his hands

media.tenor.com

Reposted by Chris Miller

Nikolai Slavov @slavov-n.bsky.social · 11d

These 3-L bottles contain one million tiny colored spheres each.

One sphere is black (1 ppm).

Finding the black sphere is comparable to detecting a protein present at ~ 6,000 copies in the proteome of a human cell.

Quantifying the protein requires analyzing multiple jars.

2 6 19

Chris Miller @chrismiller.science · 1d

I can't ever see the MTHFR gene symbol without doing a double take

3

Reposted by Chris Miller

Jeremy Berg @jeremymberg.bsky.social · 2d

I have been trying to get this published as an op-ed, but I am going to post it here since I think it is timely in light of the "consent" extortion events.

Deafening Quiet from the Scientific Establishment

jeremymberg.github.io/jeremyberg.g...

1/14

jeremymberg.github.io

13 160 300

Chris Miller @chrismiller.science · 2d

The lessons here:
1) Many gene names are stupid.
2) Edge cases may be rare, but they often matter. (TP53 is a key cancer gene that wouldn't be accessible without some special accommodations here).
3) As always, check your assumptions!
(fin)

Chris Miller @chrismiller.science · 2d

For our little internal app, this probably won't matter much, and I will either set the number of records to 200 (because we generate almost no traffic) or might code up something that dynamically decides how many queries to return, based on which genes are in the input data. (8/n)

1

Chris Miller @chrismiller.science · 2d

For those who are interested, the plot showing cumulative percentage of human HUGO gene names (from ensembl protein-coding genes) covered by a set number of records looks like this. So 8 results covers 99% of genes, 34 results covers 99.9% of genes, and it takes 199 to cover everything. (7/n)

Plot showing how many records need to be returned to ensure that each completely typed gene will be in the list.

1

Chris Miller @chrismiller.science · 2d

So in order to guarantee that we'll get "AR" in the list, the value should be 200 records, which seems excessive. My instinctual guess of 30 wasn't bad, and covers 99.89% of gene names, but that's not all of them! (6/n)

a group of pokemon standing next to each other with gotta catch 'em all written on the bottom

ALT: a group of pokemon standing next to each other with gotta catch 'em all written on the bottom

media.tenor.com

1

Chris Miller @chrismiller.science · 2d

It introduces a new question, though - this failed on TP53 with 10 results, so how many results need to be returned to handle all genes correctly? A few seconds of bash/grep later, I get the following list of 21 genes that will still fail. (5/n)

199 AR
120 PC
100 KL
78 ZNF7
67 ZNF2
67 CS
58 CP
58 ADA
57 SI
55 ZNF3
52 TH
51 C2
43 MAG
42 ZNF8
42 TNF
41 GPR1
37 DEFB1
36 USP1
36 GAL
34 PLEK
31 MET

1

Chris Miller @chrismiller.science · 2d

After some digging, it turns out that mygene.info has a default max of 10 records returned for each query, and the first 10 hits include genes like "TP53TGS", "TP53TG3F", "TP53RK-DT", but not "TP53" itself. Adding "&size=30" to the query allows it to return 30 hits, which solved this problem (4/n)

1

Chris Miller @chrismiller.science · 2d

But when I manually tried the query string - something like mygene.info/v3/query?spe... - TP53 didn't appear in the returned json - I know that's not right! (3/n)

1

Chris Miller @chrismiller.science · 2d

Digging through the backend code, I found that the tool was storing the ENSG id as the key (sensibly!), and then using an API call to mygene.info to match them up to gene names as they are typed. Seems fine... (2/n)

1

Chris Miller @chrismiller.science · 2d

Today's little mystery involving started simply enough: I was hacking on a web tool that autocompletes gene names, and was surprised when searching for "TP53" didn't return that gene. I checked, and it was definitely in the input data, so I was left scratching my head (1/n)

a close up of a man wearing glasses and a wig .

ALT: a close up of a man wearing glasses and a wig .

media.tenor.com

1 2

Chris Miller @chrismiller.science · 2d

How do we get @ensembl.org l the infrastructure they need to not be unresponsive like three times a week? Like can we pass the hat? I'm in for twenty bucks.

Reposted by Chris Miller

Michael Baym @baym.lol · 18d

Immigrants, particularly on H1Bs, are the lifeblood of American innovation. If you wanted to hurt US competitiveness in the next century, I can think of few more effective ways than a move like this

Even when found illegal, the mere intent will have irreparably harmed our future

5 37 130

Reposted by Chris Miller

Justin Wolfers @justinwolfers.bsky.social · 18d

Critical part of the President's new $100,000 charge for H1-B visas: The Administration can also offer a $100,000 discount to any person, company, or industry that it wants. Replacing rules with arbitrary discretion.

Want visas? You know who to call and who to flatter.

740 4.8K 13K

Chris Miller @chrismiller.science · 20d

When you want to do reproducible analysis in R, some packages require you to set a RNG seed. I'm not sure I trust anyone who doesn't immediately run `set.seed(42)`

1 2 2

Reposted by Chris Miller

Bede Constantinides @bedec.bsky.social · 29d

Zstandard's --long range mode works wonders for assemblies, but needs uninterrupted single line sequences.

*AllTheBacteria 661k, multiline fasta*
gzip (pigz): 751GB
zstandard --long: 641GB (30% original size)

*Single line fasta*
gzip (pigz): 700GB
zstandard --long: 232GB (10% original size)

2 12 36

Chris Miller @chrismiller.science · Sep 3

For sure. Hybrid meetings are the worst of both worlds

2

Reposted by Chris Miller

The Ass is the father of the legs @internethippo.bsky.social · Sep 3

Everyone has a second full time job being mad at the government now

59 960 6.6K

Reposted by Chris Miller

Eric Topol @erictopol.bsky.social · Sep 2

Good news. The House of Representatives stands behind the NIH budget with no cuts.

22 150 1.1K

Reposted by Chris Miller

Senator Patty Murray @murray.senate.gov · Aug 28

These are the words of a lunatic who does not belong in government, much less as our nation's top health official.

It is dangerous to allow him to oversee ALL federal health research and public health infrastructure. It is never too late to do the right thing. Fire RFK Jr.

90 380 1.6K

Reposted by Chris Miller

Ben Langmead @benlangmead.bsky.social · Aug 26

📢🚨📢🚨 Genome Informatics deadline extended to September 8! meetings.cshl.edu/meetings.asp....
Please spread the word. If you are like me and had at least one abstract that wasn't quite ready by last week's deadline, you get another swing. See you there!

Genome Informatics

Cold Spring Harbor Laboratory Meetings & Courses -- a private, non-profit institution with research programs in cancer, neuroscience, plant biology, genomics, bioinformatics.

meetings.cshl.edu

1 19 29

Reposted by Chris Miller

Ben Langmead @benlangmead.bsky.social · Aug 22

The Genome Informatics conference (@ Cold Spring Harbor Lab, Nov 5 - 8) abstract deadline is **today**. We welcome your submissions! Topics include:
- PanGenomes
- Genome Assembly & Seq. Algos.
- Algorithmic Evo. Bio
- Single Cell & Spatial Omics
- Microbial Genomics
- AI/ML & Integrative Omics
🙏🙏🙏

6 11

Reposted by Chris Miller

Evan Spotte-Smith (they/them) @ewcspottesmith.bsky.social · Aug 8

The Journal of Open Source Software (@joss-openjournals.bsky.social‬) is looking for editors. Come join our team!

I've only been an editor for a few months, but I love working with JOSS. Our peer review process is actually collaborative, and we're Diamond Open Access.

#AcademicSky 🧪 ⚗️ #CompChem

Call for editors | Journal of Open Source Software Blog

Blog for the Journal of Open Source Software • <a href='https://joss.theoj.org'>https://joss.theoj.org</a>

blog.joss.theoj.org

18 33