Steven Ge
@StevenXGe
Followers
5,853
Following
2,319
Media
220
Statuses
2,505
Professor, developer of , , ShinyGO (), & iDEP (). Topics: AI, stats, genomics, bioinformatics
Brookings, SD
Joined July 2011
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
#ปิ่นภักดิ์EP13
• 546943 Tweets
TLP LETS GET BACK TOGETHER
• 523902 Tweets
ZNN SIAM HALLOWEEN
• 288183 Tweets
MAHALIMA 2ND DAY SA ARANETA
• 236907 Tweets
Sanchez
• 213603 Tweets
Arsenal
• 184165 Tweets
選挙速報
• 154669 Tweets
当選確実
• 134894 Tweets
Tapper
• 130455 Tweets
過半数割れ
• 125948 Tweets
Palmer
• 119363 Tweets
Liverpool
• 117269 Tweets
Saka
• 73850 Tweets
Manchester United
• 72294 Tweets
Ten Hag
• 70274 Tweets
winwin
• 63403 Tweets
West Ham
• 61315 Tweets
比例復活
• 58347 Tweets
政権交代
• 53277 Tweets
Newcastle
• 52011 Tweets
Evo Morales
• 50500 Tweets
Faker
• 42317 Tweets
#MUFC
• 41961 Tweets
埼玉14区
• 40697 Tweets
#T1WIN
• 38755 Tweets
Palace
• 35710 Tweets
Dalot
• 33682 Tweets
Man U
• 33579 Tweets
GenG
• 29599 Tweets
Garnacho
• 26130 Tweets
#WHUMUN
• 25988 Tweets
IDENTITY OUT NOW
• 25705 Tweets
#ARSLIV
• 25182 Tweets
Keria
• 17986 Tweets
Rashford
• 17255 Tweets
Michael Oliver
• 16138 Tweets
Reece James
• 14633 Tweets
Ange
• 13853 Tweets
De Ligt
• 12108 Tweets
Robertson
• 11318 Tweets
Chovy
• 10898 Tweets
Merino
• 10135 Tweets
Isak
• 10002 Tweets
Pinned Tweet
Talk to your data via AI using
Massive upgrade in v0.98:
GPT-4 enables better code,
Explains codes, results, or statistics in general,
Upload a 2nd file,
EDA reports!
Demo:
3
71
281
Happy holidays! Introducing , an AI-powered app that lets you chat with your data in English! RTutor uses Davinci (
#ChatGPT
’s sibling) to turn requests into R code, which is executed & results are shown instantly, available as a HTML report in seconds. 1/8
130
2K
7K
With , you can chat with your data in plain languages. Now upgraded to v. 0.96. Fix a bug and made GPT-4 available. Interactive plots via plotly and CanvasExpress!
17
358
1K
Our
#ChatGPT
based app can do some bioinformatics analyses, from instructions in English or a few dozen other languages. Code is nicely commented. Here it conducts generates a heatmap of top genes and uses DESeq2 to identify differentially expressed genes.
7
167
770
How did humans loose our tails?
~25 million years ago, a monkey in Africa had a random mutation due to transposable elements (TEs) - pieces of DNA that can copy themselves in the human genome.
An Alu element was inserted into the intron of the TBXT gene, producing alternative
7
173
758
Wrote an R Markdown template for comprehensive exploratory data analysis (EDA). Just swap the file and render an EDA report for any data (rows & columns). It was written for , where you can upload a file and get a report in minutes.
6
148
678
New feature for GO enrichment analysis website ShinyGO v0.741 ():
Detailed genes info with links to NCBI and Ensembl.
A fully customizable lollipop plot to visualize enrichment. Is it too fancy?
27
120
685
Cut the crap and let biologists interact with their data!
With iDEP 0.951 update , you can now easily visualize the expression of all genes on all KEGG pathways.
Just upload your data, define comparisons, and select any pathway at the KEGG tab.
11
139
577
awk, sed, sort, grep, cut, uniq ... Oh My!
These Linux programs can be combined to solve problems (especially big files) elegantly in one line of Bash script.
Found a great starting doc on Linux one-liners:
by
@strnr
Bioinformatics examples included.
4
163
532
A very comprehensive tutorial for analyzing RNA-seq data by Laboratory of Petra Bleeker at the University of Amsterdam:
5
99
385
Now you can use it as an R package. Generates a comprehensive EDA report in minutes.
Wrote an R Markdown template for comprehensive exploratory data analysis (EDA). Just swap the file and render an EDA report for any data (rows & columns). It was written for , where you can upload a file and get a report in minutes.
6
148
678
2
81
371
#Bioinformatics
colleagues! Take a quick look at the top 75 Bioconductor packages.
I discovered some I should've known about but didn't.
The ones I clicked are tied to pathway analysis.
1
101
337
GPT-4 Vision generates scripts for a RNA-Seq workflow from my handwriting. Where's my job security? Need to unionize and strike now or else.
#gpt4V
9
55
289
Want to analyze RNA-Seq data interactively yet reproducibly?
Newly refactored iDEP () is almost ready. Reproducibility features:
-Generate code & reports,
-Install locally as an R package or via Docker!
10
76
265
I am shocked! My graduate student rewrote an R function in Julia and achieved 100x speed-up! He didn't know Julia, but he did it in a couple of days. His secret? GPT-4o!
It's amazing how productive someone can be with AI.
7
32
247
PhD position this fall:
AI + bioinformatics + statistics
Looking for a PhD student to start this fall. Possible topics include improving bioinformatics tools using AI and massive databases. Flexible.
Email: Xijin.Ge
@sdstate
.edu
Apply before July 5 at:
16
120
233
Gene length is important! If you get a list of genes from your study, use ShinyGO (http//ge-lab.org/go/) to examine length distribution!
"we show that transcript length alone explains most transcriptional changes observed with aging in mice and humans."
2
40
219
Available locally as an R package , RTutor can help you conduct preliminary analysis and visualization, even if you don't have any experience with R. If you do, it can save you time by providing code as a starting point. Coding in R changed, forever. 2/8
2
38
208
ShinyGO 0.80 () live in testing mode. Now supports over 14,000 species!
Updated to Ensembl 107 (215 main, 177 metazoa, 124 plants, 33 protists, & 1 bacteria), and STRING v11.5 (14,094 species).
Just bring a gene list!
2
53
198
A very thorough review of RNA-seq:
RNA Sequencing Data: Hitchhiker's Guide to Expression Analysis via
@nuzzel
thanks
@endre_sebestyen
3
81
191
In this 10-min video, I demonstrated how to use the new version of to learn R, analyze data, interpret results, and document the entire process.
0
45
187
At
#PAG30
presenting ShinyGO, a graphical tool for GO enrichment. Workshop on Tuesday 11;50 digital tools session 3, Town & Country D.
7
33
184
Is attending science conferences worth it? Just for dissemination. I spent thousands of $ and 5 days for
#PAG30
. Ten people came to my poster, 30 listened to my talk. Taking $0 and 1 min, the tweet below had 14,000 impressions, 100s of clicks. It's nice to learn & meet people.
At
#PAG30
presenting ShinyGO, a graphical tool for GO enrichment. Workshop on Tuesday 11;50 digital tools session 3, Town & Country D.
7
33
184
9
14
175
Our R01 to redevelop iDEP (), a web application for analyzing RNA-seq data. It took 1 revision, two years, and 200 support letters! These enthusiastic letters were essential. Thank you! We will make it even better! Recruiting Ph.D. students.
18
15
159
Introducing , an app (like Code Interpreter) that can greatly simplify data analysis workflows. Based on the tree-of-thoughts prompting strategy, it breaks down problems into smaller tasks, each solved through trial and error. Video:
1
53
161
Results from GO enrichment are often dominated by one type of terms, often the large ones if you sort by P-value.
If you include more terms (top50) and organize them like this, you can find that there are groups of terms related to G1/S & G2/M transition, and response to
4
23
154
Sanity checks are important in analyzing RNA-Seq data, even at read-counts level.
This example:
- total read counts vary greatly,
- rRNA removal is not uniform,
- Confirmed by # of mitochondria reads.
Now you can generate such QC plots in iDEP 1.11
5
29
149
is my holiday present to the world, which has treated me kindly. I can do things I like w/ people I like, while living well. An entirely a side project, aside from my job, done in spare times last 10 days after I learned about ChatGPT from my Tweeps. Thx
3
8
152
With , you can analyze your data by chatting. It is designed to solve complex problems through the tree-of-thoughts prompting strategy. Compared to Code Interpreter and RTutor.
1
21
146
See a cool chart on a paper?
Just copy the image to GPT-4 and ask for code!
Here I stole an idea and implemented a new way to visualize GO enrichment in ShinyGO ().
3
30
148
visNetwork is a powerful package for network visualization.
As it supports Shiny, I will use it in the ShinyGO enrichment analysis tool ().
It is also fun to play with. Maybe someone can write a computer game using R and Shiny?
5
62
145
This 9-min tutorial on took me 2 hrs to record and 5 hrs to edit.
When I grow up, I do not want to be a YouTuber... ...
0
30
141
Did you know? R cannot handle integers larger than 3 billion, which can impact popular packages such as DESeq2. R still uses 32-bit integers, and the largest allowable integer is 2^32 −1. In an unusual case, a user's RNA-Seq counts matrix included a count of 4 billion for a
4
23
134
A detailed, latest workshop on RNA-seq analysis from UC Davis.
1
25
121
Top 50 most popular tools for pathway analysis.
1. Overrepresentation analysis (ORA) using a gene list
2. Functional class scoring (FCS) using expression data or some metrics.
Based on a survey by
@mdziemann
's group
3
41
115
A comprehensive review paper (), and blog on R packages for exploratory analysis (EDA). Some auto generate reports.
Top 4 packages:
1. summarytools
2. DataExplorer
3. visdat
4. funModeling
Deep Exploratory Data Analysis (EDA) in R
2
24
115
GSVA (Gene Set Variation Analysis) is a method for pathway analysis. It translate an expression matrix into a pathway activity matrix for all pathway across samples.
This blog post by Saksham Malhotra nicely explained how GSVA works.
3
32
109
Happy to see ShinyGO is listed with tools like STRINGdb!
Hello everyone, I have made this poster with 9 open source (online) bioinformatics tools. Below I share the links to access each tool.
#Bioinformatics
#datascience
#Biology
#DNA
#genetic
#metabolomics
#neuroscience
#genomics
#proteomics
#gene
#immunology
#opensource
#tools
7
492
2K
2
28
104
Free workshop Tuesday 6pm (In-person or via Zoom)
A gentle introduction to R programming
This is a hands-on workshop for anyone interested in the R programming language. Faculty, staff, graduate, and undergraduate students are all welcome. Intended for beginners who have no or
1
34
101
Code Interpreter for Bioinformatics: Are We There Yet?
By
@hugangqing
, me & others
We tested this new
#ChatGPT
feature:
- Great for manipulating and visualizing data
- Cannot download annotation or install packages
- Concerns on storage, security, etc.
4
37
98
I checked all the brackets, nothing is wrong. But still error.
#ChatGPT
to the rescue. Costed me $0.001. Saved me 10 minutes.
3
10
93
Now published.
ShinyGO is an intuitive, graphical web application for enrichement analysis that can help researchers gain actionable insights from gene lists.
3
32
86
A laptop, a cup of tea, free YouTube music, and tons of free code online. What else does a man need?
15
5
84
Clean up column names in R:
#Rstats
1. make.names(), native in R
2. clean_names() in the janitor package
3. clean_names() in the tidyverse package
8
17
81
The depressing thing about technology is that cool, exciting innovations happen every day. Run Shiny apps without a server. Executable & modifiable R code in Quarto! If you use RStudio, watch this 6-min video by
@melvanbussel
.
0
16
73
I wanted to be a scientist.
But found myself doing software development.
Does helping other scientists count?
9
1
70
ShinyGO users, please send us an email to support our NIH grant proposal by May 31. Your email will be greatly appreciated by a small lab. We are committed to make bioinformatics tools freely accessible to all biologists, esp. those with less resources. PIs, postdoc & students!
ShinyGO 0.80 () live in testing mode. Now supports over 14,000 species!
Updated to Ensembl 107 (215 main, 177 metazoa, 124 plants, 33 protists, & 1 bacteria), and STRING v11.5 (14,094 species).
Just bring a gene list!
2
53
198
2
32
69
Not all BLAST matches are created equal.
A chart for explaining the possibilities in undergrad bioinformatics class.
0
12
66
Like a real tutor, can also explain statistics concepts and help you decided which statistical test to use. Just try the "Ask Me Anything" function!
RTutor is like an undergrad intern with some statistics and R classes! Not perfect, but helpful. 4/8
2
4
65
Besides OpenAI's amazing work, the magic is based on ingredients: your Q&As on StackOverflow, R code you shared (~20k R packages!), online books, Wikipedia, and so on. The R community is uniquely open & collaborative! Something like this is inevitable. Congratulations to all! 7/8
1
6
66
iDEP users, send us an email to support our grant proposal. Our funding expires this year. Help us by sharing your research area and how it benefits you. We need letters of support from both academic and industry users. Email GeLabinfo
@gmail
.com before Fri. 4/5. Thanks!
0
38
63
Interestingly, this preprint shows that longer genes tend to be downregulated with aging.
So, pay attention to gene length in omics studies. Our enrichment tool ShinyGO () analyzes gene length distribution in a gene list.
1
13
62
The reason I can write quickly is that I learned a lot from my students like
@ETulowetzke
in the last couple of years when we developed large bioinformatics apps like iDEP for analyzing RNA-Seq & ShinyGO for gene enrichment.
9
3
60
iDEP is a website that makes the analysis of RNA-Seq data insanely easy. To improve
#Reproducibility
, now it can generate R and R Markdown codes that can be run stand-alone.
Yes! You click around, and R code () is generated for you! It is my
#midLifeCriss
3
28
58
Just watched Intro to Python video by
@davidjmalan
As someone who mostly uses R, I always think Python is weird. This video helped.
Harvard CS50 - Lecture 6 - Python
0
11
61
Docker containers 🐳are reusable computing environments great for reproducible research. Below is the material from Ramp Krispin, an experienced data scientist at Apple, who conducted a 4-hr workshop at the SDSU Data Science Symposium last week.
The workshop focuses on the
0
13
61
Given the mpg dataset, you can ask questions like:
- “Are SUVs more fuel efficient than compact cars?”
- “Conduct ANOVA of log-transformed hwy by class and drv.”
- "Show me how to do model-based clustering."
Yes, it can be ambiguous and high level! It makes small decisions.3/8
2
1
57
SPEED2 is a web app and R package for pathway analysis based on a large collection of curated expression profiles.
1
18
58
Happy to see iDEP introduced in a bioinformatics recipe book for biologists. In Japanese. Thanks prof.
@bonohu
the editor
2
10
57
Perhaps, another reason RTutor is so impressive is the movement in R towards natural languages, i.e. the works of
@hadleywickham
and others on the ggplot2, dyplr series. RTutor is based on
@posit_pbc
's Shiny. It is a small project built on the shoulder's of giants! Thanks! 8/8
1
3
56
If you want to learn be good at developing Shiny apps, here is a hands-on approach:
1. Write a simple app w/ data you like,
2. Improve app,
3. Publish app and code (GitHub),
4. Restructure w/ Shiny Modules,
5. Make it into an R package.
1
4
56
Releasing the first part (1/4) of a video on Coding with ChatGPT. I demonstrated how to code with ChatGPT, GPT-4 with Advanced Data Analysis, , , and GitHub Copilot.
1
8
54
Gene expression atlas of a developing tissue by single cell expression correlation analysis - Nature via
@nuzzel
0
20
53
The legendary
@xieyihui
, who wrote many popular R packages (RMarkdown, bookdown, blogdown) is down. Laid off by Posit PBC. Plse join me in sponsoring him on GitHub (). The R community is unique. Even with a few $, let's show him that you care.
0
20
52
Saw this Japanese saying: If there are no failures in your life, your life might be a failure.
1
3
49
TFW one of the project ideas I always wanted to pursue "when I have time" is published.
Identifying transposable element expression dynamics and heterogeneity during development at the single-cell…
2
14
50
How to generate plots in
#RStudio
Shiny that look good on-screen & have a high resolution for publication? Here is the result of my 5 hours of googling and trying. Contributed a little bit to StackOverFlow, for the first time, after using it for 10 yrs!
0
9
48
New features for RNA-seq analysis tool iDEP ()
1. By default enrichment analysis is done using background defined by genes passed the filter.
2. Interactive network enabling users to easily visualize the relatedness of significant pathways EnrichmentMap
1
2
48
updated to v0.4:
1. Enabled https. A bit slower, but secure;
2. Interactive plots by converting ggplot2 objects or by asking Davinci to create plots using plotly;
3. Voice narration, glitchy
Once in a generation winter storm! Nothing else to do.
Cheers!
2
5
47
understands context and the generated code is even documented. 5/8
1
3
44
Hope my boss is not reading this. 😀 GitHub Copilot is so good as a VS Code plugin. I was writing some code to retrieve data from UCSC and it did 80% of the work! It even thought of things you routinely should do and generates the comments first, before writing code.
0
4
46
Another day, another slightly improved version of ShinyGO (), a graphical tool for GO enrichment analysis.
1. An option to hide GO terms that basically mean the same thing. "cell cycle" vs. "cell cycle process".
2. Filter extremely large or small pathways.
1
2
46
Go to BED, an R package for gene ID conversion
2
8
46
Interpret pathway analysis results with caution. ”pathway [analysis] methods do not inform about the activity of the pathway of interest, but instead the downstream effects of changes in the activities of transcription factors"
@AMathelier
0
17
43
When the same request is presented repeatedly, especially with a higher "temperature", finds a variety of solutions (some do not work). When asked repeatedly, "Create a correlation map of all the columns that contain numbers." 6/8
1
1
41
rearrvisr: an R package to detect, classify, and visualize genome rearrangements - via
@nuzzel
0
15
43
Deeply, deeply moved by the 130 support letters for our ShinyGO grant proposal.
Also read some of the exciting papers people published using our tool.
Such a rewarding experience, even for an reluctant developer like me.
Power to all biologist! You can analyze your own data!
1
1
42
The as.factor( ) in R creates an ordered factor. On the other hand, factor() creates an unordered factor, which you can set reference levels with relevel( ). It took me so long to learn this.
0
4
40
RTutor is a Shiny app that can be installed locally as an R package. Source code:
0
7
40
This is a minimal example of how to host large Shiny apps that can handle many users.
Based on Docker, it uses the Nginx load balancer to assign users to multiple clones of the free Shiny server.
Worked well for us for hosting iDEP, ShinyGO & RTutor.
0
4
40
Got a gene list? Try ShinyGO for enrichment analysis.
Comprehensive pathway DB for 208 plant and animal species
Highlights overlapping gene sets using trees and networks
Statistics (GC content, UTR length, etc.)
Protein Interaction Network via STRING API
6
15
38
New version of iDEP is available for testing at
Generates R code and HTML reports
Richly annotated, interactive heatmaps! You can select to zoom in and conduct enrichment analysis!
Powered by
@jokergoo_gu
's complexHeatmap!
2
4
40
Shiny from
@Rstudio
is an extremely efficient framework for developing tools. With <300 lines of code, iDEP-reads enables biologists to easily download ~18K processed RNA-seq datasets for 9 model organisms. Based on
@AviMaayan
's ARCHS4 &
@mdziemann
's DEE2
1
10
35
Ubuntu on Windows for computational biology protocol (method, recipe) by James Lloyd - via
@nuzzel
0
19
36
Topconfects: a package for confident effect sizes in differential expression analysis provides a more biologically useful ranked gene list via
@nuzzel
0
12
38
Teaching DESeq2 again this week. Read the original paper. It was elegantly written and makes so much more sense than the PowerPoint files from the internet, including those from the developers (
@mikelove
).
Read the original paper, people. Not just the abstract or tutorial.
1
0
37
What? Shiny Apps can be converted into desktop applications?
The R community keeps inventing/developing new things, way beyond a simple statistical computing language.
Shiny, Markdown, Blogdown, bookdown ... ...
What's next?
The method by
@travishinkelman
in this post works for converting
#rstats
#shiny
code into a desktop app
Deploying a
#Shiny
app as a desktop application with
#Electron
3
9
30
2
9
37
SPEAQeasy: a Scalable Pipeline for Expression Analysis and Quantification for R/Bioconductor-powered RNA-seq analyses via
@nuzzel
2
6
36
Our iDEP paper is finally out. In addition to the source code in Github, all the supporting databases are available on Zenodo.
iDEP: an integrated web application for differential expression and pathway analysis of RNA-Seq data
#bmcbioinformatics
2
13
34
This blog demonstrates "a million" ways to generate Venn Diagrams using these packages:
1. ggven
2. ggVennDiagram
3. VennDiagram
4. gplots
1
6
35
Clust : automatic extraction of optimal co-expressed gene clusters from gene expression data via
@nuzzel
0
14
34
I was finally forced to try the Linux program awk. But I have a CSV file with 200+ million lines, too much for R. With awk, I was able to quickly extract some info from each line! Wow!
5
0
31
For complex RNA-Seq expts, interpreting DESeq2 results is tricky. Always set reference levels for factors using relevel! (Details , thanks,
@mikelove
) Updated iDEP enables users to use DESeq2 via a dynamic GUI
2
12
33