peng ken lim

Public RNA-seq is a goldmine — but most of it is buried in noise. TEA-GCN just turned 450,000+ transcriptomes into a high-confidence plant gene co-expression map. Not just “more data.” A new way to extract hidden biology from the chaos. Paper: nature.com/articles/s4146… Code: github.com/pengkenlim/TEA…

大量のRNAデータから遺伝子の働きのつながりを見つける手法TEA-GCNを開発。データの偏りの影響を抑えつつ精度を向上し、遺伝子の機能予測や調節関係の解明で従来法を上回った。さらに条件ごとの特徴も捉えられ、異なる生物種間の比較にも役立つ。Nat. Commun nature.com/articles/s4146…

New in Nature Communications: TEA-GCN — a method for building gene co-expression networks from messy public RNA-seq data, without manual sample curation or batch correction. 🔗 github.com/pengkenlim/TEA… nature.com/articles/s4146… 🧵👇

Constructing gene co-functional and co-regulatory networks from public transcriptomes using condition-specific ensemble co-expression by @LabMutwil nature.com/articles/s4146…

@BVerwaaijen @LabMutwil 2nd question: We do not have a hard threshold as the true agreeabilty among the randomly generated assemblies is uncertain. The threshold is set to retain around 1/10th of total sequences from 10 assemblies so as to optimize precision without losing recall.

@LabMutwil Very interesting, does this include some level of core-genome/pan-genome identification? Which percentage of the input datasets need to support a transcript for it to be classified as true?

We developed LSTrAP-denovo, a pipeline for generating transcriptome atlases for species without a genome. All you have to do is to enter the tax ID of your species of interest... #transcriptome #assembly #CDS

@BVerwaaijen @LabMutwil Interesting question! This uses only RNA-seq data to generate transcripts. As such, we would not get any genomic information beyond coding sequences. However, since public RNA-seq data can have diverse genotypes, the final trancripts can be considered a form of pan-trancriptome.