# ![@tangming2005 Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::433559451.png) @tangming2005 Ming "Tommy" Tang

Ming "Tommy" Tang posts on X about ai, if you, shell, science the most. They currently have [------] followers and [----] posts still getting attention that total [-----] engagements in the last [--] hours.

### Engagements: [-----] [#](/creator/twitter::433559451/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::433559451/c:line/m:interactions.svg)

- [--] Week [------] +89%
- [--] Month [-------] -35%
- [--] Months [-------] +82%
- [--] Year [-------] -29%

### Mentions: [--] [#](/creator/twitter::433559451/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::433559451/c:line/m:posts_active.svg)

- [--] Week [--] +59%
- [--] Month [---] +16%
- [--] Months [---] +186%
- [--] Year [---] +323%

### Followers: [------] [#](/creator/twitter::433559451/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::433559451/c:line/m:followers.svg)

- [--] Week [------] +0.27%
- [--] Month [------] +0.83%
- [--] Months [------] +7.50%
- [--] Year [------] +20%

### CreatorRank: [-------] [#](/creator/twitter::433559451/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::433559451/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[technology brands](/list/technology-brands)  [finance](/list/finance)  [social networks](/list/social-networks)  [stocks](/list/stocks)  [countries](/list/countries)  [travel destinations](/list/travel-destinations)  [gaming](/list/gaming)  [vc firms](/list/vc-firms) 

**Social topic influence**
[ai](/topic/ai), [if you](/topic/if-you), [shell](/topic/shell), [science](/topic/science), [command](/topic/command), [how to](/topic/how-to), [youtube](/topic/youtube), [in the](/topic/in-the), [files](/topic/files), [$file](/topic/$file)

**Top accounts mentioned or mentioned by**
[@angelosgeo](/creator/undefined) [@chatomicsplaylists](/creator/undefined) [@joshuafalkenphd](/creator/undefined) [@wendy_wu123](/creator/undefined) [@chunyangbao](/creator/undefined) [@timothykassis](/creator/undefined) [@venusinadidas](/creator/undefined) [@borischernypostdtbvlmikpcmimborisandicreatedclaudecodelotsofpeoplehaveaskedhowiuseclaude](/creator/undefined) [@claudescientificskills](/creator/undefined) [@yuriybaglaenko](/creator/undefined) [@wendywu123](/creator/undefined) [@valentinesvenssonp159305684](/creator/undefined) [@jeffmold](/creator/undefined) [@tyleralban1](/creator/undefined) [@samuelmarsh](/creator/undefined) [@arjunrajlab](/creator/undefined) [@delaneyksull](/creator/undefined) [@sanyiikoooromo](/creator/undefined) [@matthewnb](/creator/undefined) [@fdmts](/creator/undefined)

**Top assets mentioned**
[FilesCoins Power Cu (FILECOIN)](/topic/files) [Alphabet Inc Class A (GOOGL)](/topic/$googl) [UiPath, Inc. (PATH)](/topic/$path) [Merge (MERGE)](/topic/merge) [UserTesting Inc. (USER)](/topic/$user)
### Top Social Posts
Top posts by engagements in the last [--] hours

"BERMUDA: a novel deep transfer learning method for single-cell RNA sequencing batch correction reveals hidden high-resolution cellular subtypes https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1764-6 https://genomebiology.biomedcentral.com/articles/10.1186/s13059-019-1764-6"  
[X Link](https://x.com/tangming2005/status/1161280532287651840)  2019-08-13T14:17Z 42.3K followers, [--] engagements


"when using kallisto bus what's the most likely reason when you get this error "no reads pseudoaligned""  
[X Link](https://x.com/tangming2005/status/1227291778337738752)  2020-02-11T18:02Z 43.2K followers, [--] engagements


"@ChunyangBao what's that"  
[X Link](https://x.com/tangming2005/status/1453115141080469505)  2021-10-26T21:43Z 43.2K followers, [--] engagements


"chatomics The best way to learn is to work through a toy dataset Generative AI: Text generation using Long short-term memory (LSTM) model https://divingintogeneticsandgenomics.com/post/generative-ai-text-generation-using-long-short-term-memory-lstm-model/ https://divingintogeneticsandgenomics.com/post/generative-ai-text-generation-using-long-short-term-memory-lstm-model/"  
[X Link](https://x.com/tangming2005/status/1800894581355286669)  2024-06-12T14:15Z 43K followers, [----] engagements


"chatomics Scanpy and Seurat marker gene log2Fold change has a big discrepancy Do you understand log2Fold change in single-cell RNAseq data https://divingintogeneticsandgenomics.com/post/do-you-really-understand-log2fold-change-in-single-cell-rnaseq-data/ https://divingintogeneticsandgenomics.com/post/do-you-really-understand-log2fold-change-in-single-cell-rnaseq-data/"  
[X Link](https://x.com/tangming2005/status/1865399622010851402)  2024-12-07T14:15Z 41.4K followers, 14.3K engagements


"1/ Three unix one-liners to turn a fastq file to a fasta file 🧵"  
[X Link](https://x.com/tangming2005/status/1886795616380182570)  2025-02-04T15:15Z 36.2K followers, [----] engagements


"Speed Up Genomics Data Processing with xargs & parallel 1/ Bioinformatics involves handling huge datasets. But are you processing them efficiently Lets talk about xargs and paralleltwo Unix commands that can massively speed up your workflows"  
[X Link](https://x.com/tangming2005/status/1889687154441531541)  2025-02-12T14:45Z 36.3K followers, 11.4K engagements


"How to Quickly Inspect Dataframe Headers in UNIX 🧵 1/ Before analyzing a big dataset you need to check the header: ✅ What variables are present ✅ What column index corresponds to each variable Heres how to do it efficiently using UNIX commands. 👇"  
[X Link](https://x.com/tangming2005/status/1890041998251557273)  2025-02-13T14:15Z 36.2K followers, [----] engagements


"UNIX File Permissions Explained 🧵 1/ Ever tried running a script and got "Permission denied" Understanding file permissions in UNIX is crucial to controlling access. Lets break it down. 👇"  
[X Link](https://x.com/tangming2005/status/1890766772439331152)  2025-02-15T14:15Z 36.3K followers, [----] engagements


"I have first-hand experience with it for over a decade. I learned Bioinformatics from scratch. It is NOT easy. How I set myself apart is by documenting what I learned"  
[X Link](https://x.com/tangming2005/status/1892216334408421859)  2025-02-19T14:15Z 36.4K followers, [---] engagements


"If you look at my github page. There are repos that I have been updating for [--] years. I take notes on what I learned"  
[X Link](https://x.com/tangming2005/status/1892216346135662774)  2025-02-19T14:15Z 36.4K followers, [---] engagements


"Now if I want to write something I just go to my github repo and find things I want to write"  
[X Link](https://x.com/tangming2005/status/1892216357657456682)  2025-02-19T14:15Z 36.4K followers, [---] engagements


"Artificial Intelligence in Medicine II https://zitniklab.hms.harvard.edu/AIM2/ https://zitniklab.hms.harvard.edu/AIM2/"  
[X Link](https://x.com/tangming2005/status/1895492916581462479)  2025-02-28T15:15Z 41.6K followers, [----] engagements


"1/ Why Bash string manipulation When processing multiple files in loops (for loop or xargs) renaming and extracting parts of filenames is crucial. Example: Convert foo.txt foo.pdf file=foo.txt echo "$file/txt/pdf" # foo.pdf"  
[X Link](https://x.com/tangming2005/status/1899449001579339805)  2025-03-11T13:15Z 42.3K followers, [---] engagements


"2/ Using basename (well-known method) To change the file extension: echo "$(basename $file .txt).pdf" # foo.pdf But theres a better way using Bash built-ins"  
[X Link](https://x.com/tangming2005/status/1899449013801525318)  2025-03-11T13:15Z 41.7K followers, [---] engagements


"2/ You have two files: A FASTA file A list of info to add to each sequence Now you want to combine them"  
[X Link](https://x.com/tangming2005/status/1919025488103866799)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"3/ Example FASTA (myfasta.txt): Blap_contig79 MSTDVDAK. Bluc_contig23663 MSTNVDAK"  
[X Link](https://x.com/tangming2005/status/1919025499780878551)  2025-05-04T13:45Z 41.7K followers, [---] engagements


"4/ Example info file (my_info.txt): info1 info2 info3 info4 You want to turn the FASTA headers into: info1_Blap_contig79"  
[X Link](https://x.com/tangming2005/status/1919025511365562666)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"6/ Whats going on here Lets break it down: paste - -: joins every two lines (header + sequence) cut -c2-: removes the leading from header"  
[X Link](https://x.com/tangming2005/status/1919025534786556065)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"7/ Then we paste my_info.txt into the header + sequence. Each line becomes: info1 Blap_contig79 MSTDVDAK"  
[X Link](https://x.com/tangming2005/status/1919025546685743288)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"8/ Finally awk formats it into FASTA again: awk 'printf("%s_%sn%sn"$1$2$3);' It rebuilds the header with the info and prints the sequence"  
[X Link](https://x.com/tangming2005/status/1919025558488572398)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"9/ Final output: info1_Blap_contig79 MSTDVDAK. info2_Bluc_contig23663 MSTNVDAK. Clean. Fast"  
[X Link](https://x.com/tangming2005/status/1919025570354229514)  2025-05-04T13:45Z 42.6K followers, [---] engagements


"9/ But here's the trap: In heatmap() if you set scale="row" clustering happens before scaling. So scaling only changes the colors. Not the clusters. That messes with your interpretation"  
[X Link](https://x.com/tangming2005/status/1921192362979787126)  2025-05-10T13:15Z 41.4K followers, [----] engagements


"10/ Want real control Use your own distance metric. Try correlation instead of Euclidean: hc - hclust(as.dist(1 - cor(t(mat)))) heatmap(mat Rowv=as.dendrogram(hc) scale="row") That gives both correct clustering and readable color range"  
[X Link](https://x.com/tangming2005/status/1921192374811980266)  2025-05-10T13:15Z 41.4K followers, [----] engagements


"Youre analyzing 10x Genomics single-cell RNA-seq and notice lots of intronic reads. Waitwasnt this a [--] UMI-based assay for mature mRNA Lets unpack why introns show upand why they matter. 🧵"  
[X Link](https://x.com/tangming2005/status/1931346667732934791)  2025-06-07T13:45Z 43.1K followers, 28.1K engagements


"3/ Imagine [-----] jellybean jars. You taste each one. Even if none have your favorite flavor you'll think [---] dojust by chance. Thats the false positive problem"  
[X Link](https://x.com/tangming2005/status/1936782507065909628)  2025-06-22T13:45Z 41.4K followers, [----] engagements


"4/ You could go strict: Bonferroni correction. Divide [----] by [-----]. Now only shout YES for jellybeans with unicorns on the label. Safebut youll miss the real ones too"  
[X Link](https://x.com/tangming2005/status/1936782518973562995)  2025-06-22T13:45Z 41.4K followers, [----] engagements


"GeneRanger is a web-server application that provides access to processed data about the expression of human genes and proteins across human cell types tissues and cell lines from several atlases. https://generanger.maayanlab.cloud/en/gene/A1CFdatabase=ARCHS4 https://generanger.maayanlab.cloud/en/gene/A1CFdatabase=ARCHS4"  
[X Link](https://x.com/tangming2005/status/1940051526669803555)  2025-07-01T14:15Z 42.9K followers, 18K engagements


"12/ Whats your scariest Excel story Drop it below. We are not getting rid of excel as it is still commonly used. awareness is important"  
[X Link](https://x.com/tangming2005/status/1949473742884831362)  2025-07-27T14:15Z 41.4K followers, [---] engagements


"Driving to pick up some toys friends gave us in the same town. My 3-year old sit in the back in a carseat. "So mommy registered for my school right" She is a little excited to go to day-care for the first time. "yes Mrs. London is going to be your teacher""  
[X Link](https://x.com/tangming2005/status/1960169924221088218)  2025-08-26T02:38Z 42.4K followers, [----] engagements


"1/ Mapping quantitative data to color https://www.nature.com/articles/nmeth.2134 https://www.nature.com/articles/nmeth.2134"  
[X Link](https://x.com/tangming2005/status/1963599201692242195)  2025-09-04T13:45Z 41.8K followers, [----] engagements


"2/ Heat map from Nature Method column https://www.nature.com/articles/nmeth.1902 https://www.nature.com/articles/nmeth.1902"  
[X Link](https://x.com/tangming2005/status/1963599213625012386)  2025-09-04T13:45Z 41.8K followers, [----] engagements


"who has power The super-connector who connects people and it is an act of generosity This has changed my view of networking you are helping others to connect and that creates power Be that super-connector"  
[X Link](https://x.com/tangming2005/status/1963782203437121838)  2025-09-05T01:52Z 43K followers, [----] engagements


"2/ https://github.com/ncborcherding/scRepertoire https://github.com/ncborcherding/scRepertoire"  
[X Link](https://x.com/tangming2005/status/1965056317615386636)  2025-09-08T14:15Z 43.1K followers, [---] engagements


"3/ python package for analyzing single cell BCR/TCR data from 10x Genomics [--] solution https://sc-dandelion.readthedocs.io/en/latest/ https://sc-dandelion.readthedocs.io/en/latest/"  
[X Link](https://x.com/tangming2005/status/1965056329929945540)  2025-09-08T14:15Z 43.2K followers, [---] engagements


"Targeting ALDH16A1 mediated thioredoxin lysosomal degradation to enhance ferroptosis susceptibility in SMARCA4-deficient NSCLC https://www.nature.com/articles/s41467-025-63687-6 https://www.nature.com/articles/s41467-025-63687-6"  
[X Link](https://x.com/tangming2005/status/1966868226765451363)  2025-09-13T14:15Z 43K followers, [----] engagements


"2/ Thats when you meet . Three dots. Small. But absolutely essential to writing flexible robust R functions"  
[X Link](https://x.com/tangming2005/status/1968310258717327662)  2025-09-17T13:45Z 41.9K followers, [--] engagements


"3/ Lets say you're writing a wrapper for plot() Without . you're stuck hardcoding everything. myplot - function(x y) plot(x y)"  
[X Link](https://x.com/tangming2005/status/1968310270620737603)  2025-09-17T13:45Z 41.9K followers, [--] engagements


"4/ But now you want to pass col = "red" or pch = [--]. Use . to pass extras down: myplot - function(x y .) plot(x y .)"  
[X Link](https://x.com/tangming2005/status/1968310282608074926)  2025-09-17T13:45Z 42.1K followers, [---] engagements


"7/ Now: missing() It checks whether an argument was provided at all. Dont confuse it with is.null()they are not the same"  
[X Link](https://x.com/tangming2005/status/1968310318284812456)  2025-09-17T13:45Z 41.9K followers, [--] engagements


"8/ Example: f - function(x) if (missing(x)) "not provided" else "provided" f() "not provided" f(NULL) "provided" missing() checks presence not value"  
[X Link](https://x.com/tangming2005/status/1968310330045636854)  2025-09-17T13:45Z 41.9K followers, [--] engagements


"9/ Compare that to: f - function(x = NULL) if (is.null(x)) "NULL" else "not NULL" f() "NULL" f(NULL) "NULL" is.null() checks the value not whether it was passed. if you read single-cell Seurat's source code you will find many if (is.null(x))"  
[X Link](https://x.com/tangming2005/status/1968310342053953621)  2025-09-17T13:45Z 42.1K followers, [--] engagements


"15/ Example in Python: from sklearn.preprocessing import normalize X_norm = normalize(X) Y_norm = normalize(Y) cos_sim = X_norm @ Y_norm.T"  
[X Link](https://x.com/tangming2005/status/1969397559660855732)  2025-09-20T13:45Z 41.9K followers, [--] engagements


"Example in R: cosine_similarity - function(X Y) normalize - function(m) m / sqrt(rowSums(m2)) X_norm - normalize(X) Y_norm - normalize(Y) cos_sim - X_norm %*% t(Y_norm) return(cos_sim)"  
[X Link](https://x.com/tangming2005/status/1969397571555901876)  2025-09-20T13:45Z 41.9K followers, [--] engagements


"6/ Prefer Python Use patchworklib. Build multi-panel figures using a similar grammar to patchwork. https://github.com/ponnhide/patchworklib https://github.com/ponnhide/patchworklib"  
[X Link](https://x.com/tangming2005/status/1970477070724850082)  2025-09-23T13:15Z 42.9K followers, [--] engagements


"7/ Keep your figure script reproducible. Use ggsave() in R or plt.savefig() in Python. Avoid manual tweaks in Illustrator if possible. Version your plots with filenames like fig1_v3.png"  
[X Link](https://x.com/tangming2005/status/1970477082615652774)  2025-09-23T13:15Z 42.9K followers, [--] engagements


"7. DiagrammeR if you want to play with R https://rich-iannone.github.io/DiagrammeR/ https://rich-iannone.github.io/DiagrammeR/"  
[X Link](https://x.com/tangming2005/status/1970847027576950895)  2025-09-24T13:45Z 43.1K followers, [---] engagements


"8. Lucidchart very similar to http://draw.io https://www.lucidchart.com/pages/ http://draw.io https://www.lucidchart.com/pages/"  
[X Link](https://x.com/tangming2005/status/1970847039346139312)  2025-09-24T13:45Z 43.1K followers, [---] engagements


"Rapid epigenomic classification of acute leukemia https://www.nature.com/articles/s41588-025-02321-z https://www.nature.com/articles/s41588-025-02321-z"  
[X Link](https://x.com/tangming2005/status/1970854498097827916)  2025-09-24T14:15Z 42.2K followers, [----] engagements


"DNA methylation cooperates with genomic alterations during non-small cell lung cancer evolution https://www.nature.com/articles/s41588-025-02307-x https://www.nature.com/articles/s41588-025-02307-x"  
[X Link](https://x.com/tangming2005/status/1971941660184793162)  2025-09-27T14:15Z 41.7K followers, [----] engagements


"Amplifying mutational profiling of extracellular vesicle mRNA with SCOPE https://www.nature.com/articles/s41587-024-02426-6 https://www.nature.com/articles/s41587-024-02426-6"  
[X Link](https://x.com/tangming2005/status/1974840758617403898)  2025-10-05T14:15Z 41.7K followers, [----] engagements


"Youre analyzing 10x Genomics single-cell RNA-seq and notice lots of intronic reads. Waitwasnt this a [--] UMI-based assay for mature mRNA Lets unpack why introns show upand why they matter. 🧵"  
[X Link](https://x.com/tangming2005/status/1976637611059052592)  2025-10-10T13:15Z 43.1K followers, [----] engagements


"Dissecting the impact of transcription factor dose on cell reprogramming heterogeneity using scTF-seq Nature Genetics https://www.nature.com/articles/s41588-025-02343-7 https://www.nature.com/articles/s41588-025-02343-7"  
[X Link](https://x.com/tangming2005/status/1977015095051079922)  2025-10-11T14:15Z 42.8K followers, 12.9K engagements


"For simple 3k PBMC [--] PCs may suffice. I have worked with really complicated neuron single cell datasets using even [---] PCs"  
[X Link](https://x.com/tangming2005/status/1979174403561918655)  2025-10-17T13:15Z 41.3K followers, [--] engagements


"6/ Some use elbow plots. Others use jackstraw or permutation tests. Heres a post I wrote on that: No method is perfect. Some of it is still feeling. https://divingintogeneticsandgenomics.com/post/permute-test-for-pca-components/ https://divingintogeneticsandgenomics.com/post/permute-test-for-pca-components/"  
[X Link](https://x.com/tangming2005/status/1979174415297581492)  2025-10-17T13:15Z 41.4K followers, [--] engagements


"7/ Okayback to k. In Seurat/Scanpy default is k=20. But should you use that It depends: How many total cells How rare are the cell types"  
[X Link](https://x.com/tangming2005/status/1979174427142259052)  2025-10-17T13:15Z 41.4K followers, [--] engagements


"12/ Also interpret carefully. KNN isnt magic. It gives you proximity not truth. It builds a scaffold. You have to decorate it with biological understanding"  
[X Link](https://x.com/tangming2005/status/1979174485585727508)  2025-10-17T13:15Z 41.3K followers, [--] engagements


"13/ Key takeaways: KNN is simple in theory nuanced in practice PCA + KNN = double dose of parameter tuning (I did not even talk about the resolution parameter.) Always visualize validate and question defaults"  
[X Link](https://x.com/tangming2005/status/1979174497283616971)  2025-10-17T13:15Z 41.3K followers, [--] engagements


"hot take: AI should learn like the human brain. it sees things (computer vision) it touches things (sensing) it tastes things (electronic tongues) it hears things (audio) it smells things (digital smell) not just through large language models (LLMs)"  
[X Link](https://x.com/tangming2005/status/1979181878767923495)  2025-10-17T13:45Z 41.5K followers, 10.5K engagements


"CRISPR tools for T cells: targeting the genome epigenome and transcriptome https://www.cell.com/trends/cancer/fulltext/S2405-8033(25)00199-2 https://www.cell.com/trends/cancer/fulltext/S2405-8033(25)00199-2"  
[X Link](https://x.com/tangming2005/status/1979189450463961346)  2025-10-17T14:15Z 41.5K followers, 16.8K engagements


"12 years ago I typed my first "Hello world". It is not easy to learn bioinformatics from scratch. But it is possible if you put in effort and have a clear pathway. read my story here https://divingintogeneticsandgenomics.com/publication/2023-10-04-nature-career/ https://divingintogeneticsandgenomics.com/publication/2023-10-04-nature-career/"  
[X Link](https://x.com/tangming2005/status/1979536708866834839)  2025-10-18T13:15Z 41.4K followers, 17.7K engagements


"10/ In single-cell RNAseq p n but data is sparse due to dropouts. PCA still crucial to reduce noise and uncover cell types "  
[X Link](https://x.com/tangming2005/status/1979551914657476766)  2025-10-18T14:15Z 41.5K followers, [---] engagements


"11/ But PCA isnt perfect. Its linear and may miss nonlinear patternsuse tSNE or UMAP for those (just note the distances between points on UMAP does not mean much)"  
[X Link](https://x.com/tangming2005/status/1979551926334480517)  2025-10-18T14:15Z 41.6K followers, [---] engagements


"1/ Still using read.csv() to load big files in R Thats like trying to fill a pool with a teacup. Heres how to go 40x faster"  
[X Link](https://x.com/tangming2005/status/1979899091447685217)  2025-10-19T13:15Z 41.4K followers, 23K engagements


"3/ Lets compare the fastest ways to read data in R: data.table::fread() vroom::vroom() readr::read_csv() Binary formats"  
[X Link](https://x.com/tangming2005/status/1979899116848291925)  2025-10-19T13:15Z 41.4K followers, [----] engagements


"4/ fread() = king of speed Multi-threaded Memory efficient Loads huge CSVs in seconds Often 840x faster than read.csv() https://jozef.io/r917-fread-comparisons/ https://jozef.io/r917-fread-comparisons/"  
[X Link](https://x.com/tangming2005/status/1979899128579871174)  2025-10-19T13:15Z 41.4K followers, [----] engagements


"10/ Tips: Use fread() or vroom() for big delimited files Convert to Feather or FST for reuse Avoid read.csv() for anything 100MB"  
[X Link](https://x.com/tangming2005/status/1979899205964730498)  2025-10-19T13:15Z 41.4K followers, [---] engagements


"11/ R can handle big data. But you need the right tools. Speed up your workflowand your sanityby ditching slow readers"  
[X Link](https://x.com/tangming2005/status/1979899217792655623)  2025-10-19T13:15Z 41.4K followers, [---] engagements


"1/ Repetition in bioinformatics Repetition isnt just tedious; its an opportunity to automate and scale your work: Say you have a list of numbers (1 [--] [--] 4) and want to multiply each by [--]. In Python you could use: A for loop A map() function List comprehension"  
[X Link](https://x.com/tangming2005/status/1979906657061842982)  2025-10-19T13:45Z 41.4K followers, [--] engagements


"2/ Now lets make this practical. In single-cell RNAseq youve learned to analyze a single sample with tools like Seurat or Scanpy. But what happens when you have [--] samples with separate count matrices from GEO 🤔"  
[X Link](https://x.com/tangming2005/status/1979906668910858647)  2025-10-19T13:45Z 41.4K followers, [--] engagements


"12/ The cookbook approach cant teach that. Only experience can. Only thinking can. Matloffs whole point is: Use your brain not just your keyboard"  
[X Link](https://x.com/tangming2005/status/1980624026474279377)  2025-10-21T13:15Z 41.4K followers, [--] engagements


"13/ Takeaways: ML isnt plug-and-play Understanding memorizing Art = adaptability Be a thinker not a copier"  
[X Link](https://x.com/tangming2005/status/1980624038470054352)  2025-10-21T13:15Z 41.4K followers, [--] engagements


"4/ The better way: Use paste. It was built for this. paste -s myfile.txt Thats it. One line. No trailing tab"  
[X Link](https://x.com/tangming2005/status/1980631469858148543)  2025-10-21T13:45Z 41.4K followers, [--] engagements


"5/ Whats -s It means serial. Instead of merging columns row by row it stacks lines horizontally into one"  
[X Link](https://x.com/tangming2005/status/1980631481560232361)  2025-10-21T13:45Z 41.4K followers, [--] engagements


"7/ Need space-separated output paste -s -d' ' myfile.txt Flexible clean zero regex gymnastics"  
[X Link](https://x.com/tangming2005/status/1980631504997986493)  2025-10-21T13:45Z 41.4K followers, [--] engagements


"8/ Why does this matter Because small formatting mistakes cascade. A trailing comma can break JSON. A rogue tab wrecks a TSV"  
[X Link](https://x.com/tangming2005/status/1980631516679139539)  2025-10-21T13:45Z 41.4K followers, [--] engagements


"Evidence of off-target probe binding in the 10x Genomics Xenium v1 Human Breast Gene Expression Panel compromises accuracy of spatial transcriptomic profilinghttps://elifesciences.org/reviewed-preprints/107070"  
[X Link](https://x.com/tangming2005/status/1980816395077386629)  2025-10-22T02:00Z 43.1K followers, [----] engagements


"Analysis of cfDNA fragmentomics metrics and commercial targeted sequencing panels https://www.nature.com/articles/s41467-025-64153-z https://www.nature.com/articles/s41467-025-64153-z"  
[X Link](https://x.com/tangming2005/status/1981001447790846347)  2025-10-22T14:15Z 41.5K followers, [----] engagements


"what style you want AI to write (tidyverse or base R for example). for a function define the desired arguments and output"  
[X Link](https://x.com/tangming2005/status/1981200476524990479)  2025-10-23T03:26Z 41.5K followers, [---] engagements


"8/ Sparsity isnt just an implementation detail. It shapes how we normalize impute cluster and interpret biology. Too many zeros PCA is unstable. Too many zeros it is just biology (not inflated) read papers in my collection https://github.com/crazyhottommy/scRNAseq-analysis-notestab=readme-ov-file#zero-inflation https://github.com/crazyhottommy/scRNAseq-analysis-notestab=readme-ov-file#zero-inflation"  
[X Link](https://x.com/tangming2005/status/1981356290208387546)  2025-10-23T13:45Z 41.5K followers, [---] engagements


"9/ Tools like Scanpy or Seurat use sparse matrix formats behind the scenes. In R: Matrix::dgCMatrix In Python: scipy.sparse.csr_matrix Learn how to work with them directly. It saves memory. And your sanity"  
[X Link](https://x.com/tangming2005/status/1981356301977588050)  2025-10-23T13:45Z 41.5K followers, [---] engagements


"chatomics new youtube video: end to end RNAseq analysis. Also I will show you how to avoid repetition of your code using list column and the purrr::map() function. https://www.youtube.com/watchv=6iAqPcFrATc https://www.youtube.com/watchv=6iAqPcFrATc"  
[X Link](https://x.com/tangming2005/status/1981711041458557391)  2025-10-24T13:15Z 41.6K followers, 14.4K engagements


"8/ Still DNNs can collapse to linear models. If you remove activations or keep them too shallow thats what youre left with"  
[X Link](https://x.com/tangming2005/status/1981718667785384269)  2025-10-24T13:45Z 41.5K followers, [--] engagements


"9/ Takeaway: DNNs aren't magic. They're just stacks of math. But once you let in non-linearity they learn structure"  
[X Link](https://x.com/tangming2005/status/1981718679453900947)  2025-10-24T13:45Z 41.5K followers, [--] engagements


"9/ Heres what I believe in: ✅ Making bioinformatics accessible ✅ Bridging wet-lab and dry-lab worlds ✅ Mentoring scientists who never thought they could code"  
[X Link](https://x.com/tangming2005/status/1982435911985152310)  2025-10-26T13:15Z 41.5K followers, [--] engagements


"10/ If youre staring at your own 2GB Excel file moment Know this: You can learn to code. You can master data. You can change lives with biology and computation. Ill help you. Thats why I share bioinformatics tips and tutorials"  
[X Link](https://x.com/tangming2005/status/1982435923704127767)  2025-10-26T13:15Z 41.5K followers, [--] engagements


"Bioinformatics isnt hard because of code (with ChatGPT you can write code much faster). Its hard because of decisions. Thousands of tiny choices that shape your results. Here's a thread"  
[X Link](https://x.com/tangming2005/status/1982443361043878031)  2025-10-26T13:45Z 41.5K followers, [----] engagements


"2/ One genome reference cant serve all purposes. Want to avoid ALT contigs Use GCA_000001405.15_GRCh38_no_alt_analysis_set.fna.gz. Want to include decoys Try hs37d5. Want telomeres and centromeres Try T2T. Every use case = different choice. https://lh3.github.io/2017/11/13/which-human-reference-genome-to-use https://lh3.github.io/2017/11/13/which-human-reference-genome-to-use"  
[X Link](https://x.com/tangming2005/status/1982443385207284062)  2025-10-26T13:45Z 41.6K followers, [--] engagements


"3/ Now you ask a basic bio question: Is gene expression anti-correlated with DNA methylation Okaybut which region near the gene 1kb upstream of TSS 2kb Maybe 200bp downstream too That one choice = wildly different results"  
[X Link](https://x.com/tangming2005/status/1982443397081416102)  2025-10-26T13:45Z 41.6K followers, [--] engagements


"5/ You want the canonical transcript Define canonical. Gencode MANE RefSeq Did you know MANE (Matched Annotation from NCBI and EMBL-EBI) provides one representative transcript per gene https://www.ncbi.nlm.nih.gov/refseq/MANE/ https://www.ncbi.nlm.nih.gov/refseq/MANE/"  
[X Link](https://x.com/tangming2005/status/1982443421207060940)  2025-10-26T13:45Z 41.6K followers, [--] engagements


"6/ Lets say you settle on 1kb upstream of the canonical TSS. Cool. But what about distal CpG sites Many regulatory regions live far from the gene. So should you use chromatin interaction data to link distal methylation Suddenly it's not so simple"  
[X Link](https://x.com/tangming2005/status/1982443433064309011)  2025-10-26T13:45Z 41.6K followers, [--] engagements


"Here is how I would start learning bioinformatics. 👇 https://divingintogeneticsandgenomics.com/post/bioinfo-roadmap/ https://divingintogeneticsandgenomics.com/post/bioinfo-roadmap/"  
[X Link](https://x.com/anyuser/status/1982798222566506842)  2025-10-27T13:15Z 43.4K followers, 29.9K engagements


"7/ script_v3_final_but_really_final_use_this_one.R is not a reproducible practice. Its a cry for help"  
[X Link](https://x.com/tangming2005/status/1982805842811597202)  2025-10-27T13:45Z 41.5K followers, [--] engagements


"There are platforms like Terra ( Despite the end of the Microsoft partnership and budget cut) AnVIL DNAnexus Code Ocean Domino LatchBio watershed and more to help with reproducibility. But no platform will fix disorganized thinking. https://anvilproject.org/ https://anvilproject.org/"  
[X Link](https://x.com/tangming2005/status/1982805854543110333)  2025-10-27T13:45Z 41.5K followers, [--] engagements


"You'll be a completely different person in a year if you 👇 Spend [--] minutes to [--] hour each day reflecting on your day. Ask yourself: - What new thing did I learn - What interesting thing did I read - Am I 0.1% better than yesterday I guarantee it. #lifelesson"  
[X Link](https://x.com/tangming2005/status/1984067662906061280)  2025-10-31T01:19Z 41.8K followers, [----] engagements


"Hey all the AI x Bio companies. I have an idea. if your agent can read this paper download the data and replicate the figures I am going to subscribe. https://www.nature.com/articles/s41586-025-09686-5 https://www.nature.com/articles/s41586-025-09686-5"  
[X Link](https://x.com/tangming2005/status/1984096051259961779)  2025-10-31T03:12Z 41.7K followers, 31.9K engagements


"I am reposting this challenge for AI companies: create an agent to tidy this spreadsheet. If your AI agent can do it I will subscribe and use it"  
[X Link](https://x.com/tangming2005/status/1984255313072324748)  2025-10-31T13:45Z 41.7K followers, [----] engagements


"This is how I did it with R https://divingintogeneticsandgenomics.com/post/my-odyssey-of-obtaining-scrnaseq-metadata/ https://divingintogeneticsandgenomics.com/post/my-odyssey-of-obtaining-scrnaseq-metadata/"  
[X Link](https://x.com/tangming2005/status/1984255325672022425)  2025-10-31T13:45Z 41.6K followers, [---] engagements


"@TimothyKassis I have been using Claude code for a week now. It blows my mind. I will need to test it on this task. so far I asked it to download llma scrape audio from youtube it transcribes to text with whisper and it fine-tuned the model and the whole workflow works"  
[X Link](https://x.com/tangming2005/status/1984488108583502051)  2025-11-01T05:10Z 41.7K followers, [--] engagements


"Big pharma seduced by transcription factors again. What has changed https://www.nature.com/articles/s41587-025-02862-y https://www.nature.com/articles/s41587-025-02862-y"  
[X Link](https://x.com/tangming2005/status/1984625238873940207)  2025-11-01T14:15Z 42.2K followers, 11.4K engagements


"@venus_in_adidas if you use perplexity it gives you the right DOI most of the time"  
[X Link](https://x.com/tangming2005/status/1984792021970674129)  2025-11-02T01:17Z 41.7K followers, [--] engagements


"2/ Building skills whether in bioinformatics or anything else requires sustained effort over time. Think of it like constructing a skyscraper. You start with one floor and keep adding moreeach one taking time and focus"  
[X Link](https://x.com/tangming2005/status/1984995190474694773)  2025-11-02T14:45Z 41.7K followers, [---] engagements


"3/ Take UNIX skills for example. Its essential for bioinformatics. At first it felt tough. But after years of consistent practice I can now navigate the terminal with ease. It wasnt quick but it was worth the effort"  
[X Link](https://x.com/tangming2005/status/1984995202386493527)  2025-11-02T14:45Z 41.7K followers, [---] engagements


"4/ Consistency is the key to mastering any skill. It's not about the occasional big effort; it's about the small daily actions that compound over time"  
[X Link](https://x.com/tangming2005/status/1984995214130577870)  2025-11-02T14:45Z 41.7K followers, [---] engagements


"12:00 AM. The Choice. Kids asleep. Body exhausted. Brain foggy. I still hadn't recorded my bioinformatics tutorial video. "I'll do it tomorrow.""  
[X Link](https://x.com/tangming2005/status/1985191048239587594)  2025-11-03T03:43Z 41.7K followers, [----] engagements


"40 minutes of recording. Energy poured in. Hit stop. Loom hadn't recorded a single second. So "tomorrow" was actually "the tomorrow I already wasted." And now tonight was the real deadline"  
[X Link](https://x.com/tangming2005/status/1985191071698330043)  2025-11-03T03:43Z 41.7K followers, [--] engagements


"A line from a book flashed through my mind: "Tomorrow is a promise that may never come. The only moment you can truly control is right now." 12:00 AM became 12:40 AM. The video got made. http://youtube.com/watchv=P7kj0GLTgS4&feature=youtu.be http://youtube.com/watchv=P7kj0GLTgS4&feature=youtu.be"  
[X Link](https://x.com/tangming2005/status/1985191083496939735)  2025-11-03T03:43Z 41.7K followers, [---] engagements


"11/ Want to understand more about p-values false discoveries and multiple testing Dive into this blog post: https://divingintogeneticsandgenomics.com/post/understanding-p-value-multiple-comparisons-fdr-and-q-value/ https://divingintogeneticsandgenomics.com/post/understanding-p-value-multiple-comparisons-fdr-and-q-value/"  
[X Link](https://x.com/tangming2005/status/1985720096262930570)  2025-11-04T14:45Z 42.1K followers, [--] engagements


"12/ Key takeaways: P-values measure surprise not truth Null hypothesis is your starting point Statistical significance is a tool not a verdict Understanding stats means telling better scientific stories"  
[X Link](https://x.com/tangming2005/status/1985720108006977648)  2025-11-04T14:45Z 42.1K followers, [---] engagements


"16/ CAR-T therapy is custom warfare. We reengineer a patients own T-cells to detect markers like CD19 or BCMA. Theyre lethal in blood cancers. In solid tumors its harderbut not impossible"  
[X Link](https://x.com/tangming2005/status/1986074975166726544)  2025-11-05T14:15Z 41.7K followers, [--] engagements


"17/ But cancer adapts. Mutates. Escapes. We see resistance through: JAK/STAT mutations blocking immune signaling Upregulation of DNA repair pathways Enzyme systems that detoxify drugs"  
[X Link](https://x.com/tangming2005/status/1986074986965254651)  2025-11-05T14:15Z 41.7K followers, [---] engagements


"A pan-KRAS inhibitor and its derived degrader elicit multifaceted anti-tumor efficacy in KRAS-driven cancers https://www.cell.com/cancer-cell/fulltext/S1535-6108(25)00310-1 https://www.cell.com/cancer-cell/fulltext/S1535-6108(25)00310-1"  
[X Link](https://x.com/tangming2005/status/1986089899825803580)  2025-11-05T15:15Z 41.8K followers, [----] engagements


"1/ Three months ago you made a figure. Today someone wants you to change it. You have no idea which script created it. This momentright hereis where reproducibility dies in bioinformatics"  
[X Link](https://x.com/tangming2005/status/1986444729643639185)  2025-11-06T14:45Z 41.7K followers, [----] engagements


"2/ You open your analysis folder. [--] R scripts. [--] Python notebooks. File names like "analysis_final_v3_ACTUAL_final.R" Which one made that figure Was it before or after the reviewer comments Did you save the filtered data The archaeology begins"  
[X Link](https://x.com/tangming2005/status/1986444742373351516)  2025-11-06T14:45Z 41.7K followers, [--] engagements


"Subclonal immune evasion in non-small cell lung cancer https://www.cell.com/cancer-cell/fulltext/S1535-6108(25)00262-4 https://www.cell.com/cancer-cell/fulltext/S1535-6108(25)00262-4"  
[X Link](https://x.com/tangming2005/status/1986814675213074862)  2025-11-07T15:15Z 41.9K followers, [----] engagements


"4/ Start with data governance. Access control. Versioning. Basic security. Do it early. Fixing leaks later costs 10x more"  
[X Link](https://x.com/tangming2005/status/1987161990788444448)  2025-11-08T14:15Z 41.8K followers, [--] engagements


"5/ Cloud is greatbut only if you use it right. Define who sees what. Set folder rules. Use Google or AWS security playbooks. Theyre free and solid"  
[X Link](https://x.com/tangming2005/status/1987162002503180664)  2025-11-08T14:15Z 41.8K followers, [--] engagements


"19/ Startups die by disorganized data. Dont be one of them. Fix your foundation nowbefore the chaos scales"  
[X Link](https://x.com/tangming2005/status/1987162167188332562)  2025-11-08T14:15Z 41.8K followers, [--] engagements


"20/ Have you seen data disasters in biotech How did you fix itor not Reply and lets trade war stories"  
[X Link](https://x.com/tangming2005/status/1987162178923995313)  2025-11-08T14:15Z 41.8K followers, [---] engagements


"I nodded. I didn't even know that's how his teacher taught it. I was learning with him. A couple days later I was driving him home from piano. He started talking. Really talking"  
[X Link](https://x.com/tangming2005/status/1987288532906614821)  2025-11-08T22:38Z 41.8K followers, [---] engagements


"About what happened in class. About the laser tag party coming up. Story after story unprompted excited. We pulled into the driveway. The moon hung huge and bright above us. "Dad is it fake How you can have a full moon for two consecutive days" I laughed"  
[X Link](https://x.com/tangming2005/status/1987288544596140268)  2025-11-08T22:38Z 41.8K followers, [---] engagements


"1/ R vs Python for bioinformatics Which one should you learn first Ive used both. I started with one. Heres what I learned the hard way:"  
[X Link](https://x.com/tangming2005/status/1987524341668647419)  2025-11-09T14:15Z 41.9K followers, 19.1K engagements


"13/ So which should you learn first If youre a biologist doing analysisR is your friend. If you want to build toolsstart with Python"  
[X Link](https://x.com/tangming2005/status/1987524483792715840)  2025-11-09T14:15Z 41.8K followers, [---] engagements


"14/ Learn one well then pick up the other. Syntax changes. But the logic Loops conditions data structurestheyre universal"  
[X Link](https://x.com/tangming2005/status/1987524495683486070)  2025-11-09T14:15Z 41.8K followers, [---] engagements


"9/ Dont get me wrong. I love deep learning. But knowing when to trust it Thats what real bioinformaticians are trained for"  
[X Link](https://x.com/tangming2005/status/1987531985821335770)  2025-11-09T14:45Z 41.8K followers, [---] engagements


"10/ Build your foundation. Then add AI on top. Thats not old school. Thats how you stop your pipeline from becoming a house of cards"  
[X Link](https://x.com/tangming2005/status/1987531997473063009)  2025-11-09T14:45Z 41.8K followers, [---] engagements


"1/ Why should every bioinformatician master Unix Because your data is just text. And Unix is the sharpest knife youll ever use. Heres why: 🧵"  
[X Link](https://x.com/tangming2005/status/1987886764770656517)  2025-11-10T14:15Z 42.2K followers, [----] engagements


"7/ Blog post with real-world examples: How I scaled from one sample to many using nothing but Unix: https://divingintogeneticsandgenomics.com/post/real-life-bioinformatics-skill-deal-with-one-sample-to-a-lot-of-samples/ https://divingintogeneticsandgenomics.com/post/real-life-bioinformatics-skill-deal-with-one-sample-to-a-lot-of-samples/"  
[X Link](https://x.com/tangming2005/status/1987886836346397075)  2025-11-10T14:15Z 42.2K followers, [---] engagements


"8/ Core Unix commands every bioinformatician should know: rev tr cut awk sed grep Theyre your Swiss Army knife. Learn them"  
[X Link](https://x.com/tangming2005/status/1987886848149176681)  2025-11-10T14:15Z 42.2K followers, [---] engagements


"10/ Learning R or Python is great. But if you skip Unix youll struggle. Unix makes you faster. Cleaner. More fearless"  
[X Link](https://x.com/tangming2005/status/1987886871733821821)  2025-11-10T14:15Z 41.8K followers, [---] engagements


"11/ Bioinformatics isnt just coding. Its craft. And Unix is part of your toolkit. Sharpen it. Use it. Own it"  
[X Link](https://x.com/tangming2005/status/1987886883486228948)  2025-11-10T14:15Z 41.8K followers, [---] engagements


"4/ First: regulatory potential scoring using exponential decay. Instead of assigning each peak to its nearest gene BETA calculates a score based on ALL nearby sites. A binding site at 5kb contributes 20x more than one at 80kb. This matches how enhancers actually work"  
[X Link](https://x.com/tangming2005/status/1987894332578505041)  2025-11-10T14:45Z 41.8K followers, [---] engagements


"5/ Second: activator or repressor prediction using the Kolmogorov-Smirnov test. If your upregulated genes cluster at the top of the regulatory potential ranking your TF is an activator. If downregulated genes cluster at top it's a repressor. This guides everything downstream"  
[X Link](https://x.com/tangming2005/status/1987894344502960428)  2025-11-10T14:45Z 41.8K followers, [---] engagements


"11/ The best computational biology tools combine elegant algorithms with real biological insight. BETA models how enhancers actually work requires evidence from multiple angles and gives you interpretable results you can validate"  
[X Link](https://x.com/tangming2005/status/1987894415843897774)  2025-11-10T14:45Z 42.1K followers, [---] engagements


"12/ Key takeaways: ChIP-seq + RNA-seq overlap is not integration Rank product requires convergent evidence Legacy tools deserve modernization AI assistants excel at translation work Resources: https://pypi.org/project/beta-binding-analysis/ https://github.com/crazyhottommy/BETA2 https://pypi.org/project/beta-binding-analysis/ https://github.com/crazyhottommy/BETA2"  
[X Link](https://x.com/tangming2005/status/1987894427722166701)  2025-11-10T14:45Z 42.1K followers, [----] engagements


""Dad which game costs the least but gets the most tickets" Birthday party at an arcade. Kids everywhere noise and flashing lights the promise of tickets you can trade for toys"  
[X Link](https://x.com/tangming2005/status/1988029986918588474)  2025-11-10T23:44Z 41.8K followers, [----] engagements


"My 7-year-old son looked around strategically. He'd figured out the game behind the games. "I need to find what costs least but gets more tickets." Smart kid. He spotted the wheel spinner. Grand prize: [---] tickets. First spin. He hit it"  
[X Link](https://x.com/tangming2005/status/1988029999207903263)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"500 tickets. Just like that. He was ecstatic. I was stunned. That kind of luck doesn't happen twice. He played some bean bag toss next - threw consistently earned [--] tickets through skill. Traded his [---] tickets for toys"  
[X Link](https://x.com/tangming2005/status/1988030011061006484)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"Then he wanted to try the wheel again. I loaded another $10. He was chasing that high. First spin: [--] tickets. Second spin: The wheel slowed. slowed. almost landed on [---] again. and clicked one space over. [--] tickets. His face fell"  
[X Link](https://x.com/tangming2005/status/1988030022893138378)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"I knelt down. "Look that 500-ticket spin That's once-in-a-lifetime luck. You can't control it. You can't repeat it." I pointed to the bean bag game"  
[X Link](https://x.com/tangming2005/status/1988030034935005253)  2025-11-10T23:44Z 41.8K followers, [--] engagements


""But that With practice you can earn [--] tickets every time. Maybe [---]. That's in your control." He looked at the wheel then at the bean bags"  
[X Link](https://x.com/tangming2005/status/1988030046687424753)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"Here's what I'm still learning as an adult: We're all chasing the 500-ticket spin. The viral post. The lucky break. The perfect timing. The big win that changes everything overnight"  
[X Link](https://x.com/tangming2005/status/1988030070091595832)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"And when we hit it once we think we've found the formula. We pour more money in. More time. More hope. Then we watch the wheel click one space over. [--] tickets. But the bean bag game That's still there"  
[X Link](https://x.com/tangming2005/status/1988030081890160806)  2025-11-10T23:44Z 41.8K followers, [--] engagements


"The boring repeatable work. The skill you can build. The consistency that compounds"  
[X Link](https://x.com/tangming2005/status/1988030093596520912)  2025-11-10T23:44Z 41.8K followers, [---] engagements


"My son earned more tickets through bean bags that day than through luck. Because luck comes once. Skill comes every time you show up. Focus on what you can control. #lifelesson"  
[X Link](https://x.com/tangming2005/status/1988030105214705960)  2025-11-10T23:44Z 41.8K followers, [---] engagements


"chatomics new blog post: how I used Claude AI to Revive BETA for Python 3: Integrating ChIP-seq and RNA-seq to Predict TF Targets https://divingintogeneticsandgenomics.com/post/reviving-beta-chip-seq-rna-seq-integration-python3/ https://divingintogeneticsandgenomics.com/post/reviving-beta-chip-seq-rna-seq-integration-python3/"  
[X Link](https://x.com/tangming2005/status/1988249125843022313)  2025-11-11T14:15Z 42.2K followers, [----] engagements


"1/ Working with big data in R Your wrangling just got a massive upgrade. duckplyr is now in the tidyverseand its fast. Really fast. 🧵"  
[X Link](https://x.com/tangming2005/status/1988256685509800078)  2025-11-11T14:45Z 42.2K followers, 16.3K engagements


"Mutation of CMTR2 in Lung Adenocarcinoma Alters RNA Alternative Splicing and Reveals Therapeutic Vulnerabilities https://www.nature.com/articles/s41467-025-64821-0 https://www.nature.com/articles/s41467-025-64821-0"  
[X Link](https://x.com/tangming2005/status/1988264218551087331)  2025-11-11T15:15Z 42.2K followers, [----] engagements


"1/ AI wont save sloppy science. Before you dive into deep learning master your foundations. Heres why basic bioinformatics still rules 🧵"  
[X Link](https://x.com/tangming2005/status/1988611508398604632)  2025-11-12T14:15Z 41.9K followers, [----] engagements


"1/ Youve got dozens of VCFs from a sequencing project. But heres the problem: Which variants are shared across multiple samples 🧵"  
[X Link](https://x.com/tangming2005/status/1988619065045504269)  2025-11-12T14:45Z 41.9K followers, [----] engagements


"2/ One line in the terminal can give you an answer. A real one. No GUI. No fancy software. Just raw Unix power"  
[X Link](https://x.com/tangming2005/status/1988619077125112116)  2025-11-12T14:45Z 41.9K followers, [---] engagements


"3/ Here it is: cat *.vcf grep -v '#' awk 'print $1 "t" $2 "t" $5' sort uniq -d wc -l Lets break it down. Line by line. Word by word"  
[X Link](https://x.com/tangming2005/status/1988619088835600876)  2025-11-12T14:45Z 41.9K followers, [---] engagements


"4/ cat *.vcf Concatenate all VCF files in the folder. Merges their contents into one stream. But were just getting started"  
[X Link](https://x.com/tangming2005/status/1988619100747415992)  2025-11-12T14:45Z 41.9K followers, [---] engagements


"5/ grep -v '#' VCF files are full of metadata lines starting with #. This removes all headers and keeps only actual variant entries"  
[X Link](https://x.com/tangming2005/status/1988619112554332604)  2025-11-12T14:45Z 41.8K followers, [---] engagements


"12/ Key takeaways: VCFs are just text UNIX commands = instant analysis uniq -d = shared variant finder Dont wait for software. Use your brain and bash"  
[X Link](https://x.com/tangming2005/status/1988619196138389907)  2025-11-12T14:45Z 41.9K followers, [---] engagements


"13/ Bioinformatics isnt always shiny dashboards. Sometimes its you a terminal and one beautiful command. Keep it simple. Make it count"  
[X Link](https://x.com/tangming2005/status/1988619207857360912)  2025-11-12T14:45Z 41.9K followers, [---] engagements


"GSEA is one of the highly cited methods ( 50K ) for gene set enrichment analysis. Do you really understand it"  
[X Link](https://x.com/tangming2005/status/1988973902505341313)  2025-11-13T14:15Z 42.2K followers, 24.8K engagements


"read the original PNAS paper here https://www.pnas.org/doi/10.1073/pnas.0506580102 https://www.pnas.org/doi/10.1073/pnas.0506580102"  
[X Link](https://x.com/tangming2005/status/1988973926509375710)  2025-11-13T14:15Z 41.9K followers, [----] engagements


"read my post on BETA but I also explains the KS test in GSEA https://divingintogeneticsandgenomics.com/post/reviving-beta-chip-seq-rna-seq-integration-python3/ https://divingintogeneticsandgenomics.com/post/reviving-beta-chip-seq-rna-seq-integration-python3/"  
[X Link](https://x.com/tangming2005/status/1988973938215645340)  2025-11-13T14:15Z 42K followers, [----] engagements


"5/ Need to trim both ends One command does it: sed 's/ t*//;s/ t*$//' file.txt First cleans the start then the end"  
[X Link](https://x.com/tangming2005/status/1988981494938001690)  2025-11-13T14:45Z 41.9K followers, [--] engagements


"6/ Lets explain this a bit: = start of line $ = end of line t = tab * = match [--] or more s/// = sed substitution pattern"  
[X Link](https://x.com/tangming2005/status/1988981506665246760)  2025-11-13T14:45Z 41.9K followers, [--] engagements


"8/ Why does this matter Pipelines break when headers dont match. Whitespace kills string matches in R Python awk bash"  
[X Link](https://x.com/tangming2005/status/1988981531013239114)  2025-11-13T14:45Z 41.9K followers, [--] engagements


"9/ Your code fails silently. And youll debug for [--] hours Only to discover an extra space after geneB"  
[X Link](https://x.com/tangming2005/status/1988981542715342947)  2025-11-13T14:45Z 41.9K followers, [--] engagements


"1/ Everyones building deep learning pipelines. But no ones asking: Does this data even make sense Lets talk about the basics. 🧵"  
[X Link](https://x.com/tangming2005/status/1988988993636299029)  2025-11-13T15:15Z 41.9K followers, [----] engagements


"2/ Bioinformatics isn't just AI single-cell transformers or generative models. It starts with a dirty truth: Biological data is messy"  
[X Link](https://x.com/tangming2005/status/1988989006080799179)  2025-11-13T15:15Z 41.9K followers, [--] engagements


"3/ Before ML before deep learning we had: QC Normalization Sanity checks Exploratory plots And it worked"  
[X Link](https://x.com/tangming2005/status/1988989017896153336)  2025-11-13T15:15Z 41.9K followers, [--] engagements


"8/ Curiosity beats complexity. Master these first: Cleaning EDA PCA Outlier detection Theyre your real power tools"  
[X Link](https://x.com/tangming2005/status/1988989076842901943)  2025-11-13T15:15Z 41.9K followers, [--] engagements


"9/ Once you trust your data then you can go deep: Classifiers Deep learning Foundation models But only then"  
[X Link](https://x.com/tangming2005/status/1988989088595341510)  2025-11-13T15:15Z 41.9K followers, [--] engagements


"1/ You think RNA equals protein Not always. If you're doing bioinformatics without understanding biology you're flying blind. 🧵"  
[X Link](https://x.com/tangming2005/status/1989336289611788752)  2025-11-14T14:15Z 42.2K followers, 13.4K engagements


"4/ Now IFNAR1. After IFN- stimulation mRNA drops slightly. But the protein gets degraded hard via ubiquitination. Big difference"  
[X Link](https://x.com/tangming2005/status/1989336325364088909)  2025-11-14T14:15Z 42K followers, [---] engagements


"5/ Or take HIF-1 under hypoxia: The mRNA is steady. But the protein is destroyed in normoxia stabilized in low oxygen. Biology flips the story"  
[X Link](https://x.com/tangming2005/status/1989336337108050325)  2025-11-14T14:15Z 42K followers, [---] engagements


"Two resources to learn bioinformatics yourself: [--]. Path to a free self-taught education in Bioinformatics #bioinformatics https://github.com/ossu/bioinformatics https://github.com/ossu/bioinformatics"  
[X Link](https://x.com/anyuser/status/1989343830454444505)  2025-11-14T14:45Z 43.4K followers, 28.2K engagements


"8 Resources to study Transcription factor binding enhancers and histone modification distribution [--]. ENCODE https://www.encodeproject.org/ https://www.encodeproject.org/"  
[X Link](https://x.com/anyuser/status/1989351387856286116)  2025-11-14T15:15Z 43.4K followers, 29.5K engagements


"2. The International Human Epigenome Consortium (IHEC) epigenome data portal http://epigenomesportal.ca/ihec/index.htmlas=1 http://epigenomesportal.ca/ihec/index.htmlas=1"  
[X Link](https://x.com/tangming2005/status/1989351400690876655)  2025-11-14T15:15Z 42K followers, [----] engagements


"3. Blueprint epigenome http://dcc.blueprint-epigenome.eu/#/home http://dcc.blueprint-epigenome.eu/#/home"  
[X Link](https://x.com/tangming2005/status/1989351412539777053)  2025-11-14T15:15Z 42K followers, [----] engagements


"1/ Exploratory Data Analysis (EDA) is the first step in any data analysis journey. When working with RNA-seq data one of the most commonly used techniques is Principal Component Analysis (PCA). But what exactly is PCA and why does it matter Lets break it down. 🧵👇"  
[X Link](https://x.com/tangming2005/status/1989698667838910894)  2025-11-15T14:15Z 42.2K followers, 15.5K engagements


"An introduction to Python for R users https://occasionaldivergences.com/posts/python-intro/ https://occasionaldivergences.com/posts/python-intro/"  
[X Link](https://x.com/anyuser/status/1989706216436867320)  2025-11-15T14:45Z 43.4K followers, 42K engagements


"4/ At first I turned to bioinformatics friends for guidance but they were too busy. It became clearI had to figure this out myself"  
[X Link](https://x.com/tangming2005/status/1990061092564091281)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"5/ I installed Ubuntu on my Windows machine and typed my first command into that intimidating black terminal. It felt like a foreign language"  
[X Link](https://x.com/tangming2005/status/1990061104924692835)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"6/ Coming from a point-and-click world command lines were scary and unintuitive. But necessity kept me going"  
[X Link](https://x.com/tangming2005/status/1990061116689645791)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"7/ The lab routine helped. Two-hour gel runs I practiced Unix. A 20-minute incubation Back to the terminal. Those intervals became gold"  
[X Link](https://x.com/tangming2005/status/1990061128437969119)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"13/ My biology background became a superpower. Understanding how data is generated helped me bridge the gap between wet lab and analysis"  
[X Link](https://x.com/tangming2005/status/1990061199095136634)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"14/ Key takeaways: Start small: Learn Unix commands during lab downtime. Pick the right tools: For genomics tidyverse + Bioconductor is invaluable. Your biology knowledge is a huge asset"  
[X Link](https://x.com/tangming2005/status/1990061210801483806)  2025-11-16T14:15Z 42.1K followers, [---] engagements


"Making a heatmap is an essential skill for a bioinformatician. But you probably do not understand heatmap. [--] reading resources to understand heatmap 🧵"  
[X Link](https://x.com/anyuser/status/1990068602482147528)  2025-11-16T14:45Z 43.4K followers, 39.1K engagements


"8/ Each modality is different: scATAC-seq is sparse Proteomics is noisy RNA-seq has 20K+ features Methylation may only cover 50K regions and over [--] million CpG sites"  
[X Link](https://x.com/tangming2005/status/1990423540852752428)  2025-11-17T14:15Z 42.2K followers, [---] engagements


"9/ Good methods normalize each modality learn weights or regularize smartly. MOFA2 DIABLO and weighted PCA all do this"  
[X Link](https://x.com/tangming2005/status/1990423552743612767)  2025-11-17T14:15Z 42.2K followers, [---] engagements


"Want to master bioinformatics data visualization Learn ggplot2 🧵👇"  
[X Link](https://x.com/tangming2005/status/1990430999759773743)  2025-11-17T14:45Z 42.2K followers, 21K engagements


"1/ Why ggplot2 In bioinformatics data visualization is essential. ggplot2 is the most powerful tool for it. But its not always intuitive at first"  
[X Link](https://x.com/tangming2005/status/1990431012367798357)  2025-11-17T14:45Z 42.2K followers, [---] engagements


""Dad you have to stay." Anna's voice stopped me. I turned. Her eyes were filling with tears. That innocent look. That need. I felt my chest tighten. The performance was starting soon. But here was my daughter asking me not to leave"  
[X Link](https://x.com/tangming2005/status/1990600913652998423)  2025-11-18T02:00Z 42.2K followers, [--] engagements


"I stayed. For two hours I sat on the floor making PlayDoh flowers with molds. The kids laughed. They made noise. They ate fruit and crackers. Anna pressed another mold into my hands: "Make this one Dad.""  
[X Link](https://x.com/tangming2005/status/1990600925443207481)  2025-11-18T02:00Z 42.2K followers, [--] engagements


"I missed Siqing Lu. I missed the performance I'd planned for the rare opportunity the music I'd wanted to hear"  
[X Link](https://x.com/tangming2005/status/1990600937237520586)  2025-11-18T02:00Z 42.2K followers, [--] engagements


"But I didn't miss the performance I was needed for. There are performances we buy tickets tothe ones we plan the ones that look impressive the ones we can tell people about. And then there are the performances where a 3-year-old hands you PlayDoh and says "stay.""  
[X Link](https://x.com/tangming2005/status/1990600949057073314)  2025-11-18T02:00Z 42.2K followers, [--] engagements


"1/ Raw counts Expression levels To compare gene expression we must normalize counts to remove sequencing biases. Heres how it works & why it matters.👇"  
[X Link](https://x.com/tangming2005/status/1990793395170914617)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"2/ Why normalize Raw read counts depend on: 📌 Sequencing depth 📌 Gene length 📌 RNA composition Without correction comparisons across samples are misleading"  
[X Link](https://x.com/tangming2005/status/1990793407036551373)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"3/ Key normalization methods: 🔹 CPM (Counts Per Million) 🔹 TPM (Transcripts Per Kilobase Million) 🔹 RPKM/FPKM (Reads/Fragments Per Kilobase Million) 🔹 DESeq2s Median of Ratios 🔹 EdgeRs TMM (Trimmed Mean of M-values)"  
[X Link](https://x.com/tangming2005/status/1990793418851938365)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"4/ CPM: Counts Per Million ✔ Adjusts for sequencing depth ✔ Useful for comparing the same gene across samples ❌ Not suitable for DE analysis Example: cpm_counts - edgeR::cpm(raw_counts)"  
[X Link](https://x.com/tangming2005/status/1990793430705074459)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"8/ EdgeRs TMM (Trimmed Mean of M-values) ✔ Corrects for sequencing depth RNA composition ✔ Great for both DE analysis"  
[X Link](https://x.com/tangming2005/status/1990793501479743813)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"Example: dge - edgeR::DGEList(counts=raw_counts) dge - edgeR::calcNormFactors(dge method="TMM") normalized_counts - edgeR::cpm(dge normalized.lib.sizes=TRUE) Use TMM when dealing with unequal RNA compositions"  
[X Link](https://x.com/tangming2005/status/1990793513056023021)  2025-11-18T14:45Z 42.2K followers, [---] engagements


"Why understanding biology matters in bioinformatics One big lesson: RNA and protein levels arent always correlated. If you dont know this you might draw the wrong conclusions. 🧵👇"  
[X Link](https://x.com/tangming2005/status/1991148223881413119)  2025-11-19T14:15Z 42.3K followers, [----] engagements


"1/ Why does this matter In bioinformatics you often analyze RNA-seq or proteomics data. If you only rely on one you risk missing the full picture. For example: mRNA and protein levels of the same gene can tell different stories due to regulation at multiple levels"  
[X Link](https://x.com/tangming2005/status/1991148236049121706)  2025-11-19T14:15Z 42.2K followers, [---] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing