3.1 Why is my gene overexpressed?¶

Theme: Theory | Analysis

On Day 2 you found genes that are differentially expressed in the Busia colony, and in Module 2.3 you checked whether they are overexpressed in other studies. Expression data tell us that a gene is overexpressed, but not why. Today we move from the transcriptome to the genome.

In this module we ask which genetic changes could explain overexpression of your candidate genes, and look for two of them in wild mosquitoes collected around Busia: copy number variants (CNVs) and amino acid changes. We use whole genomes from the MalariaGEN Ag3 resource for Ugandan and Kenyan An. gambiae collected between 2012 and 2019, including the Busia 2016 cohort that we will meet again in the selection atlas (Module 3.2).

You already know how to compute CNV and SNP frequencies with malariagen_data from the MalariaGEN–PAMCA course (W2-M4, W1-M4 and W7-M1). Here we use those tools to build evidence for (or against) a mechanism.

Learning objectives¶

At the end of this module you will be able to:

Explain three ways a gene can become overexpressed (gene amplification, cis-regulatory change, trans-acting change) and why coding changes are a separate mechanism.
Say which kinds of evidence tell these mechanisms apart, and which of them wild-caught genome data can provide.
Compute gene CNV frequencies and named CNV allele frequencies for your genes in Ugandan and Kenyan cohorts through time.
Compute amino acid substitution frequencies and combine them with CNV frequencies.
Combine population genomic evidence with AnoExpress fold changes into a short verdict for your gene dossier.

Setup¶

We pin the package versions used to write this module. Accessing Ag3 data needs a Google account; follow the data access instructions when you run Ag3() for the first time.

%pip install -q --no-warn-conflicts malariagen_data==15.9.0 anoexpress==0.3.3

import sys
import os
os.environ["MGEN_SHOW_PROGRESS"] = "0"  # hide progress bars; set to "1" to see them
import warnings
warnings.filterwarnings("ignore")

import numpy as np
import pandas as pd
import plotly.express as px
import plotly.io as pio
import malariagen_data
import anoexpress as xpe
pio.templates.default = "simple_white"  # course style for every Plotly figure

# Colab needs its own renderer; elsewhere, save figures as Plotly JSON so the book can show them
pio.renderers.default = "colab" if "google.colab" in sys.modules else "plotly_mimetype+notebook_connected"
pd.set_option("display.max_columns", 50)

ag3 = malariagen_data.Ag3()
ag3

# Replace with the genes you were assigned on Day 2 (see data/candidate-genes.tsv)
CANDIDATE_GENES = ["AGAP002862", "AGAP009193"]  # CYP6AA1, GSTE4

df_genes = ag3.genome_features(attributes=["ID", "Name", "description"]).query(
    "type == 'gene' and ID in @CANDIDATE_GENES"
)
df_genes[["ID", "Name", "contig", "start", "end", "strand", "description"]]

Theory: how does a gene become overexpressed?¶

A gene is transcribed more when there are more copies of it, when its own regulatory DNA changes, or when the proteins that regulate it change. Changes to the protein itself are a fourth, separate route to resistance.

1. Gene amplification (CNV). A duplication copies the gene (often with its neighbours) so the genome carries three, four or more copies. More templates usually means more transcript. Metabolic resistance genes in Anopheles are frequently amplified: Lucas et al. [1] found CNVs at the Cyp6aa/Cyp6p, Cyp6m/z, Cyp9k1 and Gste clusters across Africa, and a duplication spanning Cyp6aa1 (Cyp6aap_Dup1) has spread across East Africa together with other resistance mutations [2].

2. Cis-regulatory change. A SNP, indel or transposable element insertion in the promoter, enhancer or UTR of the gene changes how strongly that copy is transcribed. The best-understood example in malaria vectors is in An. funestus, where promoter variants and an insertion upstream of CYP6P9a/b drive overexpression [3, 4].

3. Trans-acting change. A change in a transcription factor or signalling pathway raises the expression of many genes at once, wherever they are in the genome. The Maf-S / cnc / Keap1 pathway regulates many detoxification genes in An. gambiae [5], and the selection atlas flags a sweep at Keap1 (Module 3.2).

4. Coding changes (an alternative, not an explanation). An amino acid substitution can make an enzyme better at binding or breaking down an insecticide without changing its expression, for example GSTE2 L119F in An. funestus [6]. Coding changes do not explain overexpression, but they often sit on the same haplotype as a CNV or regulatory variant and can be useful markers of it.

Finally, remember that a fold change is a comparison: a gene can look up or down because of the reference strain, rearing conditions or insecticide exposure (induction), not only because of genetic change in the resistant population.

How can we tell the mechanisms apart?¶

Evidence	CNV	cis	trans	Can we get it from wild-caught WGS?
Higher sequencing coverage over the gene; discordant reads at breakpoints	yes	no	no	yes (`gene_cnv`, `cnv_discordant_read_calls`)
Expression rises with copy number across individuals	yes	no	no	needs RNA and DNA from the same individuals
Allele-specific expression in heterozygotes	no (usually)	yes	no	needs RNA from heterozygotes
Neighbouring genes up together	often (co-amplified)	sometimes	only if co-regulated	partly (look at neighbours in the CNV)
Many unlinked genes up together	no	no	yes	no (use AnoExpress co-expression, Module 2.3)
Selection signal at the gene	yes	yes	no, it sits at the regulator	yes (Module 3.2)
Candidate variants at high frequency near the gene	CNV alleles	promoter SNPs / insertions	variants at the regulator	yes (SNP and CNV frequencies)

So with Ag3 we can test the CNV hypothesis directly, find coding changes, and find variants that rise in frequency near the gene. We cannot prove a cis or trans mechanism with genomes alone, but we can say which explanations the data support.

The cohorts: Busia and its neighbours¶

Busia (Uganda) sits on the Kenyan border. Ag3 has An. gambiae from Busia in 2016, 2017 and 2019, from neighbouring districts (Namayingo, Mayuge, Tororo), from Teso North in Busia County on the Kenyan side (2019), and from Kanungu in western Uganda (2012) for contrast. Siaya itself (Day 4) is not in Ag3; Teso North is the closest Kenyan cohort.

SAMPLE_SETS = [
    "AG1000G-UG",                     # Tororo and Kanungu, 2012
    "1178-VO-UG-LAWNICZAK-VMF00025",  # Busia 2016-17, Mayuge 2017
    "1288-VO-UG-DONNELLY-VMF00168",   # Namayingo 2017-19
    "1288-VO-UG-DONNELLY-VMF00219",   # Busia 2017 and 2019
    "1274-VO-KE-KAMAU-VMF00246",      # Teso North (Busia County, Kenya) 2019
]
DISTRICTS = ["Busia", "Namayingo", "Mayuge", "Tororo", "Teso North", "Kanungu"]
SAMPLE_QUERY = f"taxon == 'gambiae' and admin2_name in {DISTRICTS}"
COHORTS = "admin2_year"  # group by district and year; labels look like UG-E_Busia_gamb_2016

df_samples = ag3.sample_metadata(sample_sets=SAMPLE_SETS, sample_query=SAMPLE_QUERY)
df_samples.pivot_table(
    index=["country", "admin2_name"], columns="year", values="sample_id",
    aggfunc="count", fill_value=0,
)

Frequency functions drop cohorts with fewer than 10 mosquitoes by default (min_cohort_size=10), so the single Busia 2013 mosquito and the four Teso North 2013 mosquitoes are ignored.

CNVs at your candidate genes¶

gene_cnv_frequencies() summarises the copy number HMM calls by gene: for each cohort it reports the fraction of mosquitoes with more copies (amp) or fewer copies (del) than expected (see PAMCA W2-M4). We can pass gene IDs straight to region. The function also returns any other gene that overlaps the region, so we keep only our candidates.

cnv_frq = ag3.gene_cnv_frequencies(
    region=CANDIDATE_GENES,
    cohorts=COHORTS,
    sample_sets=SAMPLE_SETS,
    sample_query=SAMPLE_QUERY,
    drop_invariant=False,
).query("gene_id in @CANDIDATE_GENES")

ag3.plot_frequencies_heatmap(cnv_frq, title="Gene CNV frequencies, candidate genes");

A CNV at your gene is a candidate explanation only if it is common in the population the RNA came from. A CNV found in 2 % of mosquitoes cannot explain a large fold change in a colony unless the colony happens to carry it, so always look at the frequency near Busia.

Worked example 1: the Cyp6aa/Cyp6p duplication¶

The Cyp6aa/Cyp6p cluster on 2R (~28.48–28.51 Mb) holds Cyp6aa1, Cyp6p3, Cyp6p4 and other P450s. Let’s look at gene CNV frequencies across the whole cluster.

CYP6_REGION = "2R:28,480,000-28,510,000"

cyp6_cnv = ag3.gene_cnv_frequencies(
    region=CYP6_REGION,
    cohorts=COHORTS,
    sample_sets=SAMPLE_SETS,
    sample_query=SAMPLE_QUERY,
)
ag3.plot_frequencies_heatmap(
    cyp6_cnv.query("max_af > 0.05"), title="Gene CNV frequencies, Cyp6aa/Cyp6p cluster"
);

Almost every mosquito from Busia, Namayingo, Mayuge and Teso North carries an amplification of Cyp6aa1, whereas amplification of Cyp6p3 is found in at most 15 % of mosquitoes in any cohort. The coverage HMM tells us that a gene is amplified, not which duplication it is. For the known resistance loci, Ag3 also provides named CNV alleles called from discordant and split reads at the breakpoints [1], e.g. Cyp6aap_Dup1a. The Dup0 entries are an “any amplification at this locus” aggregate, not an allele.

The function below turns those calls into carrier frequencies per cohort, in the same format as the other frequency functions so we can reuse plot_frequencies_heatmap().

def cnv_allele_frequencies(contig, prefix, sample_query=SAMPLE_QUERY, cohorts=COHORTS,
                           min_cohort_size=10):
    # Fraction of mosquitoes in each cohort that carry each named CNV allele.
    ds = ag3.cnv_discordant_read_calls(
        contig=contig, sample_sets=SAMPLE_SETS, sample_query=sample_query
    )
    ok = ~ds["sample_is_high_variance"].values  # drop noisy samples, as gene_cnv_frequencies does
    calls = pd.DataFrame(
        ds["call_genotype"].values[:, ok].T,
        index=ds["sample_id"].values[ok],
        columns=ds["variant_id"].values,
    )
    alleles = [a for a in calls.columns if a.startswith(prefix) and not a.endswith("Dup0")]
    df_meta = ag3.sample_metadata(sample_sets=SAMPLE_SETS, sample_query=sample_query)
    cohort_col = "cohort_" + cohorts
    df = calls[alleles].join(df_meta.set_index("sample_id")[cohort_col])
    sizes = df[cohort_col].value_counts()
    df = df[df[cohort_col].isin(sizes[sizes >= min_cohort_size].index)]
    frq = (df.groupby(cohort_col)[alleles].mean().T).add_prefix("frq_")
    frq["max_af"] = frq.max(axis=1)
    frq["label"] = frq.index
    return frq.query("max_af > 0")


cyp6_alleles = cnv_allele_frequencies(contig="2R", prefix="Cyp6aap")
ag3.plot_frequencies_heatmap(cyp6_alleles, title="Cyp6aa/Cyp6p CNV alleles (carrier frequency)");

The amplification is one allele, Cyp6aap_Dup1a, a duplication that includes Cyp6aa1 but not Cyp6p3. It is close to fixation around Busia and lower in Tororo and Kanungu in 2012. This is the duplication described by Njoroge et al. [2] as part of a “triple mutant” haplotype that spread through East Africa.

Let’s plot the carrier frequency through time in each district.

def tidy_allele_frequencies(frq):
    # Long format with district and year parsed from labels like UG-E_Busia_gamb_2016.
    long = (
        frq.filter(like="frq_")
        .rename(columns=lambda c: c.removeprefix("frq_"))
        .reset_index(names="allele")
        .melt(id_vars="allele", var_name="cohort", value_name="frequency")
    )
    parts = long["cohort"].str.split("_", expand=True)
    long["district"] = parts[1].str.replace("-", " ")
    long["year"] = parts[3].astype(int)
    return long


px.line(
    tidy_allele_frequencies(cyp6_alleles).query("allele == 'Cyp6aap_Dup1a'"),
    x="year", y="frequency", color="district", markers=True, range_y=[0, 1.05],
    title="Cyp6aap_Dup1a carrier frequency", width=700, height=400,
)

Worked example 2: Cyp9k1 amplifications¶

Cyp9k1 on the X chromosome is another P450 linked to pyrethroid resistance [7] and is often amplified [1]. On X the function accounts for males having one copy.

cyp9k1_cnv = ag3.gene_cnv_frequencies(
    region="AGAP000818",
    cohorts=COHORTS,
    sample_sets=SAMPLE_SETS,
    sample_query=SAMPLE_QUERY,
).query("gene_id == 'AGAP000818'")

cyp9k1_alleles = cnv_allele_frequencies(contig="X", prefix="Cyp9k1")

ag3.plot_frequencies_heatmap(
    pd.concat([cyp9k1_cnv, cyp9k1_alleles]),
    title="Cyp9k1: gene CNV (HMM) and named CNV alleles",
);

Between a third and a half of mosquitoes around Busia carry a Cyp9k1 amplification, mostly the allele Cyp9k1_Dup8. How many extra copies do they carry? gene_cnv() gives the modal copy number over each gene for each mosquito.

BUSIA_2016 = "cohort_admin2_year == 'UG-E_Busia_gamb_2016' and sex_call == 'F'"

ds_cn = ag3.gene_cnv(region="AGAP000818", sample_sets=SAMPLE_SETS, sample_query=BUSIA_2016)
ix = list(ds_cn["gene_id"].values).index("AGAP000818")
copy_number = pd.Series(ds_cn["CN_mode"].values[ix], name="copy number")
copy_number.value_counts().sort_index().rename("n mosquitoes (Busia 2016 females)")

Females normally have 2 copies of an X-linked gene; carriers of the amplification have 3 or more. In Module 3.3 we will see that the Cyp9k1 amplification sits on a swept haplotype.

Amino acid changes in your candidate genes¶

Next we look for common amino acid substitutions with aa_allele_frequencies() (PAMCA W1-M4). We use each gene’s canonical transcript and keep changes above 5 % in at least one cohort.

aa_frqs = {}
for gene in CANDIDATE_GENES:
    transcript = ag3.canonical_transcript(gene)
    df = ag3.aa_allele_frequencies(
        transcript=transcript,
        cohorts=COHORTS,
        sample_sets=SAMPLE_SETS,
        sample_query=SAMPLE_QUERY,
    )
    aa_frqs[gene] = df.query("max_af > 0.05")
    print(f"{gene} ({transcript}): {len(aa_frqs[gene])} amino acid changes above 5 %")

AGAP002862 (AGAP002862-RA): 17 amino acid changes above 5 %
AGAP009193 (AGAP009193-RA): 12 amino acid changes above 5 %

for gene, df in aa_frqs.items():
    if len(df) > 0:
        ag3.plot_frequencies_heatmap(df, title=f"Amino acid frequencies, {gene}")

Amino acid changes that are close to fixation in every cohort (including Kanungu, far from Busia) are usually differences between the reference genome (the PEST strain) and East African An. gambiae, not resistance mutations. Look for changes that vary between places or times.

Combining SNPs and CNVs at the Cyp6 cluster¶

At the Cyp6 cluster, CYP6P4 I236M is part of the East African “triple mutant” together with Cyp6aap_Dup1 and a transposon insertion upstream of Cyp6p4 [2]. Following the pattern from PAMCA W7-M1, we put SNP and CNV frequencies on one heatmap.

cyp6p4_aa = ag3.aa_allele_frequencies(
    transcript="AGAP002867-RA",  # CYP6P4
    cohorts=COHORTS,
    sample_sets=SAMPLE_SETS,
    sample_query=SAMPLE_QUERY,
).query("max_af > 0.2")

cyp6_combined = pd.concat([cyp6p4_aa, cyp6_cnv.query("max_af > 0.2"), cyp6_alleles.query("max_af > 0.2")])
ag3.plot_frequencies_heatmap(cyp6_combined, title="Cyp6 cluster: amino acid, gene CNV and CNV allele frequencies");

Where Cyp6aap_Dup1a is common, CYP6P4 I236M is common too, which is what we expect if they travel on the same haplotype. But in Kanungu I236M is at about 80 % while the duplication is carried by only a quarter of mosquitoes, so I236M must also occur on haplotypes without the duplication. (SNP genotypes called inside an amplified region can be biased, see W7-M1; we test the linkage directly in Exercise 2, and with haplotypes in Module 3.3.)

Time series make the trend easier to see. The *_advanced functions group by area and period and add confidence intervals.

ds_cyp6p4 = ag3.aa_allele_frequencies_advanced(
    transcript="AGAP002867-RA",
    area_by="admin2_name",
    period_by="year",
    sample_sets=SAMPLE_SETS,
    sample_query=SAMPLE_QUERY,
    variant_query="max_af > 0.5",
)
ag3.plot_frequencies_time_series(ds_cyp6p4, height=600, width=900, title="CYP6P4 I236M");

Bringing in the expression data¶

Finally, we put the genomic evidence next to the expression evidence from AnoExpress (Module 2.3). We load log2 fold changes and adjusted p-values for your candidate genes and the three worked examples in the An. gambiae/An. coluzzii analysis. The Busia contrast in AnoExpress is BusiaSurvivors_v_Kisumu (deltamethrin-selected Busia colony survivors vs the susceptible Kisumu strain).

EXAMPLE_GENES = ["AGAP002862", "AGAP002865", "AGAP000818"]  # CYP6AA1, CYP6P3, CYP9K1
BUSIA_CONTRAST = "BusiaSurvivors_v_Kisumu"
genes = list(dict.fromkeys(CANDIDATE_GENES + EXAMPLE_GENES))

fc = xpe.data(data_type="fcs", analysis="gamb_colu", gene_id=genes)
pv = xpe.data(data_type="pvals", analysis="gamb_colu", gene_id=genes)

names = (
    ag3.genome_features(attributes=["ID", "Name"])
    .query("type == 'gene'")
    .drop_duplicates("ID")
    .set_index("ID")["Name"]
)
up = (fc > 0) & (pv < 0.05)
expression = pd.DataFrame({
    "name": names.reindex(fc.index),
    "busia_log2fc": fc[BUSIA_CONTRAST].round(2),
    "busia_padj": pv[BUSIA_CONTRAST].round(3),
    "median_log2fc_all": fc.median(axis=1).round(2),
    "n_contrasts_up": up.sum(axis=1),
    "n_contrasts": fc.notna().sum(axis=1),
})
expression

fc_long = fc.reset_index().melt(id_vars="GeneID", var_name="contrast", value_name="log2FC")
fc_long["name"] = fc_long["GeneID"].map(names).fillna(fc_long["GeneID"])
fc_long["study"] = np.where(fc_long["contrast"] == BUSIA_CONTRAST, "Busia", "other studies")
fig = px.strip(
    fc_long, x="name", y="log2FC", color="study", hover_data=["contrast"],
    title="AnoExpress fold changes (gamb_colu)", width=800, height=450,
)
fig.add_hline(y=0, line_dash="dot")
fig

Now we build a small evidence table: expression next to the CNV and amino acid frequencies in Busia 2016 and 2019.

all_cnv = pd.concat([cnv_frq, cyp6_cnv, cyp9k1_cnv]).reset_index()
amp = (
    all_cnv.query("cnv_type == 'amp'")
    .drop_duplicates("gene_id")
    .set_index("gene_id")[["frq_UG-E_Busia_gamb_2016", "frq_UG-E_Busia_gamb_2019"]]
    .add_prefix("amp_")
)

aa_max = {
    gene: df.filter(like="frq_UG-E_Busia").max().max() if len(df) else 0.0
    for gene, df in aa_frqs.items()
}

evidence = expression.join(amp)
evidence["max_missense_frq_busia"] = pd.Series(aa_max)
evidence.round(2)

Read the table with care:

CYP6AA1 is strongly overexpressed in many resistant populations, especially in West Africa, but it is not significantly different in BusiaSurvivors_v_Kisumu (log2FC about −0.3). The case for CYP6AA1 in Busia rests on other evidence: the near-fixed Cyp6aap_Dup1a duplication shown above, the East African “triple mutant” described by Njoroge et al. [2], and the selected-vs-parental comparison from our own RNA-Seq-Pop run (Day 2).
CYP6P3 is strongly down in Busia survivors vs Kisumu. The Kisumu samples in that experiment express CYP6P3 highly (see Module 2.3): a fold change depends on the reference strain.
A variant that is fixed in Busia cannot explain expression differences within Busia, and selection on a haplotype (Modules 3.2 and 3.3) can act on any of the mutations it carries.

For your own genes, write down which mechanism(s) the data support, which they rule out, and what experiment would settle it (for example qPCR of copy number and expression in the same mosquitoes, or allele-specific expression).

Exercise 1¶

Look at the CNV heatmap and the amino acid tables for your own candidate genes (run the notebook with your genes in CANDIDATE_GENES). For each gene:

Is it amplified in Busia? At what frequency, and is that changing between 2016 and 2019?
Are there common amino acid changes that vary between cohorts?
Which mechanism(s) of overexpression are still possible after this analysis?

Exercise 2¶

Are CYP6P4 I236M and Cyp6aap_Dup1a on the same haplotype? In Busia almost every mosquito carries the duplication, so test it in Tororo 2012 (AG1000G-UG), where both are at intermediate frequency. Cross-tabulate each mosquito’s number of 236M alleles (SNP 2R:28,497,967 G>C) against whether it carries Cyp6aap_Dup1a.

Hint: ag3.snp_calls(region="2R:28497967-28497967", ...) and ag3.cnv_discordant_read_calls(contig="2R", ...).

query = "cohort_admin2_year == 'UG-E_Tororo_gamb_2012'"
ds_snp = ag3.snp_calls(region="2R:28497967-28497967", sample_sets="AG1000G-UG", sample_query=query)
alleles = ds_snp["variant_allele"].values[0].astype(str)  # ['G', 'C', ...]
gt = ds_snp["call_genotype"].values[0]                    # (samples, ploidy)
n_236m = pd.Series((gt == list(alleles).index("C")).sum(axis=1), index=ds_snp["sample_id"].values)

ds_cnv = ag3.cnv_discordant_read_calls(contig="2R", sample_sets="AG1000G-UG", sample_query=query)
dup = pd.Series(
    ds_cnv["call_genotype"].values[list(ds_cnv["variant_id"].values).index("Cyp6aap_Dup1a")],
    index=ds_cnv["sample_id"].values,
)
pd.crosstab(n_236m.rename("n 236M alleles"), dup.rename("Cyp6aap_Dup1a carrier"))

In our run: of 39 mosquitoes homozygous for 236M, 31 carry the duplication; of 14 homozygous for 236I, only 2 do; heterozygotes are split (32 of 59). The two mutations are strongly associated, but not perfectly: many 236M haplotypes lack the duplication. This fits a stepwise history in which the duplication arose on a haplotype that already carried I236M. So a SNP assay for I236M tracks the triple-mutant haplotype well where it dominates (Busia), but would overestimate the duplication elsewhere. Module 3.3 looks for SNPs that tag the duplicated haplotype more precisely.

Exercise 3¶

Kanungu (western Uganda, 2012) is far from Busia. Compare Cyp9k1 CNVs there with the Busia cohorts. Is it the same allele? What does that suggest about how Cyp9k1 amplifications arose?

Exercise 4¶

Cyp6aap_Dup1a is almost fixed in Busia, yet CYP6AA1 is not significantly overexpressed in BusiaSurvivors_v_Kisumu and CYP6P3 is lower in Busia survivors than in Kisumu. Give two explanations, and one experiment that would distinguish them.

Quiz¶

Check your understanding. The quiz runs in Colab and in the course book.

%pip install -q jupyterquiz==2.9.6.4
from jupyterquiz import display_quiz
display_quiz("https://raw.githubusercontent.com/sanjaynagi/anopheles-omics-training/main/quizzes/3-1-cnvs-and-snps.json")

Summary¶

Overexpression can come from gene amplification, cis-regulatory change or trans-acting change; coding changes are a separate route to resistance that often travels on the same haplotype.
Wild-caught genomes test the CNV hypothesis directly (coverage and breakpoint calls) and reveal common coding and linked variants; cis vs trans needs expression data from the same individuals.
Around Busia, Cyp6aap_Dup1a is near fixation and travels with CYP6P4 I236M; Cyp9k1_Dup8 is carried by a third to a half of mosquitoes.
Genomic and expression evidence must be read together, and fold changes depend on the reference strain.

Well done!¶

Add the CNV and SNP results for your genes to your gene dossier. In Module 3.2 we ask whether your genes sit inside selective sweeps in wild populations.

References¶

Lucas ER et al. (2019). Whole-genome sequencing reveals high complexity of copy number variation at insecticide resistance loci in malaria mosquitoes. Genome Research 29:1250–1261. Lucas et al. (2019)
Njoroge H et al. (2022). Identification of a rapidly-spreading triple mutant for high-level metabolic insecticide resistance in Anopheles gambiae provides a real-time molecular diagnostic for antimalarial intervention deployment. Molecular Ecology. Njoroge et al. (2022)
Mugenzi LMJ et al. (2019). Cis-regulatory CYP6P9b P450 variants associated with loss of insecticide-treated bed net efficacy against Anopheles funestus. Nature Communications 10:4652. Mugenzi et al. (2019)
Weedall GD et al. (2019). A cytochrome P450 allele confers pyrethroid resistance on a major African malaria vector, reducing insecticide-treated bednet efficacy. Science Translational Medicine 11:eaat7386. Weedall et al. (2019)
Ingham VA et al. (2017). The transcription factor Maf-S regulates metabolic resistance to insecticides in the malaria vector Anopheles gambiae. BMC Genomics 18:669. Ingham et al. (2017)
Riveron JM et al. (2014). A single mutation in the GSTe2 gene allows tracking of metabolically based insecticide resistance in a major malaria vector. Genome Biology 15:R27. Riveron et al. (2014)
Vontas J et al. (2018). Rapid selection of a pyrethroid metabolic enzyme CYP9K1 by operational malaria control activities. PNAS 115:4619–4624. Vontas et al. (2018)
Nagi SC et al. (2023). RNA-Seq-Pop: exploiting the sequence in RNA-Seq, a Snakemake workflow reveals patterns of insecticide resistance in the malaria vector Anopheles gambiae. Molecular Ecology Resources 23:1195–1206. Nagi et al. (2023)
MalariaGEN Vector Observatory data and malariagen_data documentation: https://malariagen.github.io/vector-data/
AnoExpress: https://github.com/sanjaynagi/AnoExpress

References¶

Lucas, E. R., Miles, A., Harding, N. J., Clarkson, C. S., Lawniczak, M. K. N., Kwiatkowski, D. P., Weetman, D., & Donnelly, M. J. (2019). Whole-genome sequencing reveals high complexity of copy number variation at insecticide resistance loci in malaria mosquitoes. Genome Research, 29(8), 1250–1261. 10.1101/gr.245795.118
Njoroge, H., van’t Hof, A., Oruni, A., Pipini, D., Nagi, S. C., Lynd, A., Lucas, E. R., Tomlinson, S., Grau‐Bove, X., McDermott, D., Wat’senga, F. T., Manzambi, E. Z., Agossa, F. R., Mokuba, A., Irish, S., Kabula, B., Mbogo, C., Bargul, J., Paine, M. J. I., … Donnelly, M. J. (2022). Identification of a rapidly‐spreading triple mutant for high‐level metabolic insecticide resistance in Anopheles gambiae provides a real‐time molecular diagnostic for antimalarial intervention deployment. Molecular Ecology, 31(16), 4307–4318. 10.1111/mec.16591
Mugenzi, L. M. J., Menze, B. D., Tchouakui, M., Wondji, M. J., Irving, H., Tchoupo, M., Hearn, J., Weedall, G. D., Riveron, J. M., & Wondji, C. S. (2019). Cis-regulatory CYP6P9b P450 variants associated with loss of insecticide-treated bed net efficacy against Anopheles funestus. Nature Communications, 10(1). 10.1038/s41467-019-12686-5
Weedall, G. D., Mugenzi, L. M. J., Menze, B. D., Tchouakui, M., Ibrahim, S. S., Amvongo-Adjia, N., Irving, H., Wondji, M. J., Tchoupo, M., Djouaka, R., Riveron, J. M., & Wondji, C. S. (2019). A cytochrome P450 allele confers pyrethroid resistance on a major African malaria vector, reducing insecticide-treated bednet efficacy. Science Translational Medicine, 11(484). 10.1126/scitranslmed.aat7386
Ingham, V. A., Pignatelli, P., Moore, J. D., Wagstaff, S., & Ranson, H. (2017). The transcription factor Maf-S regulates metabolic resistance to insecticides in the malaria vector Anopheles gambiae. BMC Genomics, 18(1). 10.1186/s12864-017-4086-7
Riveron, J. M., Yunta, C., Ibrahim, S. S., Djouaka, R., Irving, H., Menze, B. D., Ismail, H. M., Hemingway, J., Ranson, H., Albert, A., & Wondji, C. S. (2014). A single mutation in the GSTe2 gene allows tracking of metabolically based insecticide resistance in a major malaria vector. Genome Biology, 15(2). 10.1186/gb-2014-15-2-r27
Vontas, J., Grigoraki, L., Morgan, J., Tsakireli, D., Fuseini, G., Segura, L., Niemczura de Carvalho, J., Nguema, R., Weetman, D., Slotman, M. A., & Hemingway, J. (2018). Rapid selection of a pyrethroid metabolic enzyme CYP9K1 by operational malaria control activities. Proceedings of the National Academy of Sciences, 115(18), 4619–4624. 10.1073/pnas.1719663115
Nagi, S. C., Oruni, A., Weetman, D., & Donnelly, M. J. (2023). RNA‐Seq‐Pop : Exploiting the sequence in <scp>RNA</scp> sequencing—A Snakemake workflow reveals patterns of insecticide resistance in the malaria vector Anopheles gambiae. Molecular Ecology Resources, 23(4), 946–961. 10.1111/1755-0998.13759