A polynomial Turing compression (PTC) for a parameterized problem $L$ is a polynomial time Turing machine that has access to an oracle for a problem $L'$ such that a polynomial in the input parameter bounds each query. Meanwhile, a polynomial (many-one) compression (PC) can be regarded as a restricted variant of PTC where the machine can query the oracle exactly once and must output the same answer as the oracle. Bodlaender et al. (ICALP 2008) and Fortnow and Santhanam (STOC 2008) initiated an impressive hardness theory for PC under the assumption coNP $\not\subseteq$ NP/poly. Since PTC is a generalization of PC, we define $\mathcal{C}$ as the set of all problems that have PTCs but have no PCs under the assumption coNP $\not\subseteq$ NP/poly. Based on the hardness theory for PC, Fernau et al. (STACS 2009) found the first problem Leaf Out-tree($k$) in $\mathcal{C}$. However, very little is known about $\mathcal{C}$, as only a dozen problems were shown to belong to the complexity class in the last ten years. Several problems are open, for example, whether CNF-SAT($n$) and $k$-path are in $\mathcal{C}$, and novel ideas are required to better understand the fundamental differences between PTCs and PCs. In this paper, we enrich our knowledge about $\mathcal{C}$ by showing that several problems parameterized by modular-width ($mw$) belong to $\mathcal{C}$. More specifically, exploiting the properties of the well-studied structural graph parameter $mw$, we demonstrate 17 problems parameterized by $mw$ are in $\mathcal{C}$, such as Chromatic Number($mw$) and Hamiltonian Cycle($mw$). In addition, we develop a general recipe to prove the existence of PTCs for a large class of problems, including our 17 problems.