Colorectal cancer (CRC) is the most common gastrointestinal malignancy worldwide. The poor specificity and sensitivity of the fecal occult blood test has prompted the development of CRC-related genetic markers for CRC screening and treatment. Gene expression profiles in stool specimens are effective, sensitive and clinically applicable. Herein, a novel advantage of using cells shed from the colon is presented for cost-effective CRC screening. Molecular panels were generated through a series of leave-one-out cross-validation and discriminant analyses. A logistic regression model following reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and immunohistochemistry was used to validate a specific panel for CRC prediction. The panel, consisting of ubiquitin-conjugating enzyme E2 N (UBE2N), inosine monophosphate dehydrogenase 1 (IMPDH1), dynein cytoplasmic 1 light intermediate chain 1 (DYNC1LI1) and phospholipase A and acyltransferase 2 (HRASLS2), accurately recognized patients with CRC and could thus be further investigated as a potential prognostic and predictive biomarker for CRC. UBE2N, IMPDH1 and DYNC1LI1 expression levels were upregulated and HRASLS2 expression was downregulated in CRC tissues. The predictive power of the panel was 96.6% [95% confidence interval (CI), 88.1-99.6%] sensitivity and 89.7% (95% CI, 72.6-97.8%) specificity at a predicted cut-off value at 0.540, suggesting that this four-gene panel testing of stool specimens can faithfully mirror the state of the colon. On the whole, the present study demonstrates that screening for CRC or cancer detection in stool specimens collected non-invasively does not require the inclusion of an excessive number of genes, and colonic defects can be identified via the detection of an aberrant protein in the mucosa or submucosa.