Macroautophagy/autophagy can selectively degrade misfolded proteins, damaged organelles and other cargoes. It is conceivable that alteration of the degradation processes could disrupt normal cellular signaling and contribute to human diseases such as cancer. To explore the link between aberrant autophagy selectivity and human cancer, we have developed a pipeline called "inference of cancer-associated LC3-interacting region-containing proteins" (iCAL), which integrates a sequence-based predictor, a model-based computational method, publicly available cancer mutations, and multiple experimental approaches. Using iCAL, we have identified 222 LIR motif-associated mutations (LAMs) in 148 LIR-containing proteins (LIRCPs), and validated that LAMs in ATG4B, STBD1, EHMT2 and BRAF impair their interactions with LC3 and/or autophagy activities. Moreover, we uncovered that STBD1, a previously poorly-characterized protein, inhibits tumor growth via metabolism reprogramming in cancer cells. A patient-derived mutation in STBD1 (W203C) disrupts the interaction with LC3 and promotes tumor growth. Taken together, iCAL provides an exciting new avenue to discover novel autophagy pathways that contribute to carcinogenesis.