“With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.” — John von Neumann, Mathematician (1953)

In 2023, a trio of computational biologists achieved something mathematically brilliant but biologically absurd: they took a highly complex, multi-dimensional dataset of single-cell RNA sequencing and projected it into the exact shape of a walking elephant [1].

They did not do this because the cells biologically resembled an animal. They did it to issue a profound warning to the scientific community. By tweaking the initialization parameters of Picasso, a custom algorithm based on popular dimensionality-reduction tools like UMAP, they demonstrated that data could be warped into virtually any arbitrary shape—all while maintaining the standard statistical metrics researchers use to “prove” their data is valid. (A similar interactive essay by Google PAIR previously showed how tweaking UMAP could morph a 3D dataset of a mammoth skeleton into unrecognizable, abstract blobs [2]).

These visual demonstrations are microcosms of a much larger philosophical and practical crisis in modern science. Bioinformatics and biostatistics are simplified mathematical models used to interpret an unfathomably complex biological reality. But what happens when scientists mistake the mathematical model for the biological truth?

The result is a cascade of statistical illusions, a replication crisis that has rocked top-tier scientific journals, and billions of dollars wasted in pharmaceutical development. Yet, despite these profound flaws, medical science still manages to produce miracles. To understand why, we must examine the friction between the rigidity of mathematics, the messiness of biology, and the unforgiving filter of physical reality.

🌐 Other languages: English 日本語

Have you ever run into this situation while using the ΔΔCt method for qPCR analysis—your control group’s mean expression value isn’t exactly 1? What did you do? Did you quietly normalize the ΔΔCt results again just to force the control group’s mean back to 1? And have you ever wondered whether this “sneaky extra normalization” is actually valid, or if you’ve been doing something questionable all along?

Well, in this blog post, we’re going to get to the bottom of it—is this secondary normalization mathematically justified, or is it just a placebo for our inner perfectionist? Let’s find out.

This document explains how we calculate the contribution score for BSGOU (Bioinformatics Study Group in Okayama University) members. It is designed to fairly distribute recognition or shared resources (e.g., donations) based on actual contributions, while being robust against gaming by spammy commits or shallow participation.

Some things in this world just don’t make sense—no matter how you slice it. Take the zebrafish, for example. A tiny, transparent freeloader that wouldn’t even pick it as bait, much less any fishing dude’s Instagram story. And yet—somehow—it ended up becoming a pillar of modern life sciences. Yeah, I know. It sounds ridiculous. So, how did this happen? Let’s rewind.

1. A Tiny Life, Barely Noticed

Do you know where zebrafish originally come from?

I made a silly story this week…

🌐 Other languages: English 日本語中文

1. Background

RNA modifications are well-documented contributors to post-transcriptional regulation. Among them, glycoRNAs—RNAs conjugated with glycans—are a recently described and poorly understood class of biomolecules. Initial evidence for their existence was reported in 2021 by Flynn et al., who detected glycosylated small RNAs (Ryan, et al., Cell, 2021). However, due to a lack of suitable imaging methods, their precise cellular distribution and potential functional roles remained unclear.

To address this, Ma et al. (Yuan Ma, et al., Nat Biotechnol, 2024) developed an in situ imaging method termed ARPLA (Aptamer and RNA in situ hybridization-mediated Proximity Ligation Assay). This method aims to detect glycoRNAs with high spatial resolution, selectivity, and sequence specificity.

Presentation1