Biological
Computational
Experimental
Discussions
Fun Facts
100

What is an enhancer and what does it control?

A cis-regulatory DNA element whose sequence encodes cell-type/temporal control of transcription via TF motifs and their grammar.

100

What does the deep learning do in this study?

Scores sequences (and every possible SNV) for a cell-type class to guide design (evolution or motif placement).

100

How were fly designs tested in vivo?

Genomic integration of enhancer + minP–nGFP; adult brains imaged for cell-type-specific fluorescence (KC/PNG).

100

How quickly can random DNA become a cell-type enhancer?

In ~10–15 SNVs, with scores saturating and in vivo activity appearing.

100

Give two pieces of evidence that mitochondria (or chloroplasts) originated from bacteria.

Any two from: circular genomes; 70S-type ribosomes; binary-fission-like replication; double membranes with bacterial lipids; sensitivity to some antibiotics; phylogenies placing them with α-proteobacteria (or cyanobacteria for chloroplasts).

200

What do the authors mean by motif grammar?

The rules for which TF motifs occur and their spacing, orientation, order, and clustering that together specify output.

200

Summarize the in silico evolution algorithm.

Greedy hill-climb: at each round test all SNVs, keep the best-scoring mutation, iterate ~10–15 steps.

200

What two assays were used in human cells and what do they read out?

Episomal minP–luciferase (activity strength) and genomic integration + ATAC-seq (chromatin accessibility gain/loss).

200

What spacing rules did KC motif implantation reveal?

Mef2 ~5 bp upstream of Ey; Onecut ~3 bp downstream of Ey (strand/order matter).

200

During an all-out 30–60 s sprint, what actually causes muscle acidification, and what happens to lactate afterward?

The immediate acid load comes mainly from ATP hydrolysis releasing H⁺, not from lactate per se. Lactate is oxidized as fuel or recycled in the Cori cycle to glucose in the liver.

300

Name the key activators and a repressor in each system.

Fly γ-KC: Ey, Mef2, Onecut (±Sr); Human MEL: SOX10, MITF, TFAP2; 

Repressor: ZEB (human) and sequence-specific silencer motifs in fly.

300

How does motif implantation differ from evolution?

It pre-specifies motifs (vocabulary) and uses the oracle to place them at best positions/strands/spacings (syntax), instead of discovering grammar via stepwise edits.

300

Why use a minimal promoter (minP) and include a random-drift control?

minP isolates enhancer-driven output; random drift shows increases require directed editing (random walks stay near baseline).

300

What is a proto-enhancer and how was it demonstrated?

An initially inactive genomic fragment that becomes functional after ~6–11 SNVs; authors “rescue” such loci to KC activity.

300

Humans inherit mitochondrial DNA almost exclusively from mothers. What is the proximal cellular mechanism that removes paternal mtDNA after fertilization?

Ubiquitin-tagging and selective mitophagy of sperm mitochondria after fertilization; plus massive egg-to-sperm mtDNA copy-number imbalance and physical exclusion of sperm midpieces.

400

What prior observations justify the claim that regulatory capacity is in the enhancer DNA sequence?

Short DNA fragments drive cell-type-specific reporter activity; sequence changes predictably alter activity; high-throughput assays show sequence variation explains quantitative enhancer output.

400

Propose an experiment to generate more cell-type-specific epigenetic labels to train DeepFlyBrain.

Perform cell-type-resolved CUT&Tag/CUT&RUN (or ChIP-seq) for H3K27ac/H3K4me1 and key TFs (Ey/Mef2/Onecut) in sorted fly brain cell types; integrate these tracks as multi-task targets (accessibility + marks + TF occupancy).

400

How did the authors validate computationally designed enhancers in both species?

Fly: integrated minP–nGFP reporters showing KC/PNG-specific patterns. Human: luciferase strength and integrated ATAC peaks that appear with evolved designs and disappear after repressor addition.

400

In Drosophila, (a) how did introducing repressor sites affect scores/activators? (b) Did spacing/order among activators matter?

(a) Scores collapsed and activity switched off even with activators present (repression dominates). (b) Yes—bp-scale syntax (Ey–Mef2 ≈5 bp; Ey–Onecut ≈3 bp; correct order/strand) was required for high scores.

400

In CRISPR–Cas9, briefly state the roles of spacer acquisition, the crRNA:tracrRNA (or sgRNA), and the PAM sequence.

Spacer acquisition (Cas1/Cas2) captures invader DNA into the CRISPR array (adaptive memory); crRNA:tracrRNA/sgRNA guides Cas9 to the matching DNA; PAM is a short motif required for Cas9 binding/cleavage and helps prevent self-targeting.

500

What does “multiple cell-type codes” mean, and what in-paper examples show it?

One enhancer can carry two programs: it can be expanded (amon: T4 → T4+KC) or restricted (Pkc53e: KC+T → KC-only) by ~10 SNVs.

500

How do attribution and Δ-prediction (in silico saturation mutagenesis) guide edit choice?

They localize bases whose change gives the largest score drop and often creates repressor motifs near activator clusters—prioritizing causal SNVs.

500

How were repressor effects causally tested and why is that informative?

By adding a few repressor-creating SNVs near activator clusters and re-assaying; scores, reporter signals, and ATAC drop, proving repression dominance and giving a programmable off-switch.

500

What general rules and transferability did the human section confirm?

SOX10-centered heterotypic modules (±MITF, TFAP2) follow stereotyped placements; short edit paths or rule-based implantation create MEL-specific enhancers; ZEB provides an off-switch; predictions agree across DeepMEL2 ↔ luciferase/ATAC ↔ Enformer/ChromBPNet.

500

Large, long-lived animals don’t have sky-high cancer rates. Propose at least two mechanistic solutions evolution has used.

Mechanisms include expanded tumor-suppressor dosage or activity, enhanced DNA repair / genome maintenance, lower per-cell proliferation rates, early contact inhibition / extracellular matrix constraints, etc.