Genetic Screens, Genome Editing, and Sequencing Methods

Mutagenesis + screen

Finds all genes needed for a process. Mutagenize, plate single cells as colonies, test each colony for the phenotype (e.g. dye stains dead autophagy mutants). Hits are mutant colonies. Limited by how many colonies you can check.

Mutagenesis + selection

Finds rare mutants (e.g. drug resistance). Plate under conditions where wild type dies and only mutants of interest grow. Survivors are the hits. Very high throughput.

Auxotroph screen (Beadle & Tatum)

Maps steps in a biosynthetic pathway. Irradiate Neurospora, find mutants that fail on minimal medium but grow with one added nutrient, then test them on pathway intermediates. The mutant rescued by the most supplements is blocked earliest.

Patch plating (lac screens)

Finds uninducible and constitutive mutants. Copy colonies to glycerol, then lactose (uninducible = grows on glycerol only) or glycerol + X-gal (constitutive = blue without lactose).

Plaque screen (lambda)

Finds lysogeny and lysis genes. Plate phage on a bacterial lawn and look at plaque shape. Cloudy = wild type. Clear = can't lysogenize (cI, cII, cIII mutants).

Su(var) / E(var) screens

Finds heterochromatin factors. Mutagenize variegating flies and look for second mutations that change the red/white pattern. More red = Su(var). More white = E(var).

Tetrad dissection

Tests whether a trait is a single gene. Cross mutant × wild type, sporulate the diploid, separate the four spores of each ascus, score each. 2:2 = single gene. Other ratios = more than one gene.

Complementation test (yeast)

Counts how many genes a set of mutants hit. Mate two haploid recessive mutants and score the diploid. Wild type = different genes. Mutant = same gene.

Merodiploid / F' complementation (E. coli)

Tests cis vs trans behavior in bacteria. Put some alleles on an F' plasmid and others on the chromosome, then measure enzyme with and without inducer. Dominant in cis only = cis element. Same in cis and trans = diffusible protein.

Deletion construct (goi::DrugR)

Knocks out a known gene (works well in yeast). DrugR gene flanked by homology to the gene's neighbors, transform, select on drug, verify by PCR. Survivors replaced the gene by HDR. In mouse most survivors are wrong integrations.

Yeast deletion library with barcodes

Genome-wide knockout phenotyping. One strain per gene, each with UPTAG and DNTAG barcodes. Pool strains, grow under a condition, sequence barcodes. A barcode that disappears = gene needed in that condition.

CRISPR-Cas9 knockout

Makes a gene-disrupting indel in any organism. sgRNA guides Cas9 to a target next to NGG, Cas9 cuts both strands, NHEJ repairs with indels, frameshift disrupts the gene.

CRISPR-Cas9 + donor DNA

Makes precise point mutations, knock-ins, deletions. Same as knockout but also supply donor DNA with homology arms carrying the change. HDR copies the donor into the break.

Pooled CRISPR screen (Wang 2015)

Finds essential genes in cultured cells. Virus library with one guide RNA per gene, one gene disrupted per cell, grow many generations, count remaining guides. Guides that vanish targeted essential genes.

RNAi

Temporary knockdown, works in non-model organisms. Introduce dsRNA matching the gene and RISC destroys the matching mRNA. Phenocopies loss of function. Not heritable, no DNA change.

Cre/lox conditional knockout

Studies an essential gene in one tissue or time. Flank the gene with same-orientation loxP sites (floxed) and drive Cre from a tissue-specific promoter (e.g. Alb-Cre). Cre excises the gene only where expressed. Control: floxed without Cre is normal.

Gal4-LexA domain swap

Shows TF domains are modular. Fuse DNA-binding and activation domains from different proteins and measure lacZ reporter. LexA-BD + Gal4-AD activates at LexA sites.

lacZ reporter in transgenic embryo

Tests whether DNA is an enhancer (sufficiency). Put candidate upstream of lacZ in a transgenic mouse and stain with X-gal. Blue only in the tissue where the enhancer works (limb buds for Shh).

Enhancer deletion or point mutation

Tests whether an enhancer is necessary. Delete or mutate it in the animal, or find point mutations in patients. Loss of expression and phenotype (short limbs for Shh enhancer).

ChIP / ChIP-seq

Shows where a TF or histone mark sits on DNA. Cross-link proteins to DNA, fragment, pull down with antibody, reverse cross-links, PCR or sequence the DNA. Reads mapped to the genome show binding or mark locations.

Bisulfite sequencing

Shows which cytosines are methylated. Bisulfite turns unmethylated C into U (reads as T) and methylated C resists. A C that stays C = methylated.

DNase protection (footprinting)

Defines a protein's DNA binding site (e.g. Lac operator). Bound protein protects its DNA from DNase and the protected fragments are sequenced. Protected sequence = binding site.

Purified protein binding assay

Shows direct binding in a test tube. Mix purified repressor with operator DNA with and without IPTG (radioactive IPTG tracks it). Binds operator, released by inducer.

Yeast two-hybrid

Tests whether two proteins interact. Gal4 BD fused to bait, Gal4 AD fused to target, in yeast with a Gal4-driven reporter. Reporter ON = interaction reconstituted Gal4.

RT-PCR

Measures one mRNA (e.g. over time). Reverse transcribe RNA to cDNA then amplify. IFN-β mRNA first appeared about 6 h after infection.

Radioactive amino acids (Monod)

Shows induction makes new enzyme. Add label before vs after inducer. Radioactivity ends up in the new enzyme molecules.

Sanger sequencing

Accurate reads of small pieces (600-1000 bp). Polymerase extends a primer with fluorescent chain-terminating dideoxy nucleotides. Color order gives sequence.

Illumina (NGS)

Genomes, RNA-seq, ChIP-seq, exomes. Shear DNA, add adaptors, bridge amplification makes clusters on a flow cell, add one fluorescent base per cycle for 100-300 cycles. Color sequence per cluster = read.

PacBio / Nanopore

Very long reads for repeats. Read single DNA molecules directly (10 kb to 4 Mb). Easier assembly but more errors.

Shotgun sequencing + assembly

Builds a genome. Sequence random fragments and a computer merges overlaps into contigs. Overlapping reads vote so errors average out.

Paired-end reads

Join contigs into scaffolds. Sequence both ends of a fragment of known size. Ends in different contigs = neighbors at about that distance.

Clone / chromosome walking

Old, accurate, slow sequencing. Sequence a stretch, design a primer from its end, repeat.

Exome sequencing

Finds disease mutations cheaply. Capture exons with biotin probes and sequence only 30-60 Mb. Finds de novo mutations in rare disease or cancer.

mRNA-seq

Exon structure and expression level. Extract RNA, oligo-dT beads pull mRNA, fragment to about 300 bases, reverse transcribe, PCR adds barcode, sequence, map to genome. Reads pile on exons, count ≈ expression, missing exon = alternative splicing.

cDNA / ESTs

Proves gene structure. Align cDNA to genomic DNA. Aligned = exons, gaps = introns. ESTs mark transcript ends.

ORF detection

Finds candidate genes in raw sequence. Scan both strands in all three frames (six frames) for runs of codons with start and stop.

BLAST

Guesses function. Search for similar sequences in other organisms. A similar known gene suggests similar function.

Comparative genomics (conservation)

Finds functional sequence including non-coding. Align related genomes and look for conserved regions. Coding: 3-nt periodicity, frame-preserving gaps, splice sites. Conserved with no coding pattern: suspect enhancer or RNA gene.

Which genes are needed for a process I can score?

Forward genetic screen (mutagenesis + screen)

Wild type dies but mutants of interest grow

Genetic selection

Each colony must be tested to find mutants

Genetic screen

Is a trait caused by a single gene?

Cross to wild type and dissect tetrads (2:2)

Tetrads show 1:3 or 0:4 instead of 2:2

More than one gene is involved

Do two recessive mutants hit the same gene?

Complementation test (mate them, score the diploid)

Diploid from two recessive mutants is wild type

Mutations are in different genes

Diploid from two recessive mutants is mutant

Mutations are in the same gene

How do you do complementation in haploid E. coli?

Merodiploids built on F' plasmids

What does my favorite gene do?

Knockout (deletion construct or CRISPR) or RNAi

I need a precise point mutation in a mouse

Cas9 + sgRNA + donor DNA with homology arms (HDR)

I need a quick gene disruption with Cas9

Cas9 + sgRNA and let NHEJ make indels

Which genes are essential in cultured human cells?

Pooled CRISPR guide-RNA screen

Which genes does yeast need under low phosphate?

Pooled barcoded deletion library, sequence barcodes

Gene is essential but I want its role in the liver

Cre/lox with a liver promoter (Alb-Cre)

I want to knock down a gene in a non-model organism without changing DNA

RNAi (dsRNA)

Which genes are on in a sample and how much?

mRNA-seq (read count ≈ expression)

Reads cover exons 1, 2, 4 in kidney but all four in brain

Alternative splicing (exon 3 skipped in kidney), seen by RNA-seq

Where does a TF or histone mark sit in the genome?

ChIP-seq

Which cytosines are methylated?

Bisulfite sequencing

Which histone tags were at a promoter over time?

ChIP (IFN-β study)

Do two proteins physically interact?

Yeast two-hybrid

Is a TF domain modular?

Domain swap (Gal4 AD + LexA BD) with a reporter

What DNA sequence does a repressor protect?

DNase protection (footprinting)

Does induction make new enzyme or activate old enzyme?

Radioactive amino acids before vs after inducer

Is a conserved non-coding region an enhancer?

Conservation, then lacZ reporter in a transgenic embryo (sufficiency)

Is that enhancer necessary?

Delete or mutate it and look for loss of expression

Reporter is blue only in limb buds

The element is a limb enhancer (sufficient to drive expression there)

How do you find disease mutations in a rare-disease child cheaply?

Exome sequencing (look for de novo mutations)

How do you build a genome from small reads?

Shotgun sequencing and computer assembly

How do you join contigs into scaffolds?

Paired-end reads of known fragment size

How do you get through repeats?

Long reads (PacBio / Nanopore) or long-insert paired ends

Which sequencing method gives very high throughput with short reads?

Illumina

Which sequencing method gives 600-1000 bp accurate reads?

Sanger

How do you decide if a sequence is a gene in a new genome?

ORF detection, cDNA/RNA-seq, conservation, BLAST

How do you find heterochromatin factors?

Su(var) / E(var) screens in variegating flies

Which pathway order does a set of auxotrophs give?

Test each mutant on intermediates; most supplements rescue = earliest block

Mutant colonies are clear plaques on a phage lawn

Phage can't lysogenize (cI, cII, cIII)

Mutant is blue on glycerol + X-gal

Constitutive lac mutant

Mutant grows on glycerol but not lactose

Uninducible lac mutant

Which method proves a gene's function?

Reverse genetics (disrupt the gene); omics data are only suggestive