*2.5. Sequencing and Bioinformatic Analysis*

Extracted DNA samples were sent to Macrogen for library preparation with TruSeq DNA PCR-Free kit and sequenced using 151 bp paired-end Illumina platform. We obtained 185,173,048 fragments for the ML10 wild-type pool, 173,761,432 for the M5 mutant, 207,190,254 for the pool of F2 wild-type individuals, 235,111,914 for the pool of F2 mutant individuals. Reads were mapped to the B73 reference genome (*Zea mays*, Ensemble release 40, http://plants.ensembl.org) using bwa mem [33] and further processed using samtools [34]. Variant calling was performed using bcftools [35,36]. Data analyses were done using R (R Core Team, 2017. https://www.R-project.org/), plots were generated using R/lattice [37] (http://lmdvr.r-forge.r-project.org). Variant callings with quality scores below 500 were discarded. Allele frequencies (ratio of sample over reference B73) for F2 pools were plotted for variants fixed between the E1-9 mutant sample (allele frequency equal to 1) and the ML10 wild-type sample (allele frequency equal to 0) and having an allele frequency between 0.25 and 0.75 in the F2 wild-type pool sample. Loess smoothing for average allele frequencies was plotted to visualize more general tendencies. Coding sequences (CDS) for known fasciation genes were extracted manually using IGV [38]. Sequences were translated to peptides using transeq and then multiple sequences aligned using muscle [39].

Sequencing data are available at NCBI SRA under accession number PRJNA602200.
