Fig. 4.

Relationship between LN–LN ($\mathbf{M}$) and ORNs $\to$ LN ($\mathbf{W}$) synaptic counts in the connectome reconstruction. (A) LN–LN connections synaptic counts $\mathbf{M}$ on the left and right sides of the larva. (B) ${\mathbf{W}}^{\text{T}}\mathbf{W}$ with $W=[w_{\text{LN1},\mathrm{...},}{w}_{\text{LN8}}]$ on the left and right sides. Thus each entry is $w_{\text{LNi}}^{\text{T}}{w}_{\text{LNj}}$, the scalar product between 2 ORNs $\to$ LN synaptic count vectors $w_{\text{LN}}$. (C) ${({\mathbf{W}}^{\text{T}}\mathbf{W})}^{1/2}$, i.e., the square root of the matrices in (B). (D) Entries of $\mathbf{M}$ vs entries of ${({\mathbf{W}}^{\text{T}}\mathbf{W})}^{1/2}$, excluding the diagonal, for both sides. r: Pearson correlation coefficient. pv: one-sided P-value calculated by shuffling the entries of each $w_{\text{LN}}$ independently, which assures that each LN keeps the same total number of synapses. Shuffling the entries of $\mathbf{M}$ in addition to shuffling each $w_{\text{LN}}$ leads to P-value $<{10}^{-4}$.