Purpose　Colorectal　cancer　(CRC)　is　one　of　the　most　common　and　fatal　gastrointestinal　malignancies,　in　which　cancer　stem　cells　(CSCs)　were　identified　to　enable　tumor　heterogeneity　and　initiate　tumor　formation.　However,　the　process　from　CSCs　to　invasion　cells　is　unconfirmed.　Methods　Several　bioinformatics　methods,　including　clustering,　pseudotime　analysis,　gene　set　variation　analysis　and　gene　ontology　enrichment,　were　used　to　construct　a　path　of　gradual　transformation　of　CSCs　to　invasive　cells,　called　＂stem-to-invasion　path＂.　A　large　amount　of　signaling　interactions　were　collated　to　build　the　multilayer　regulatory　network.　Kaplan-Meier　curve　and　time-dependent　ROC　method　were　applied　to　reveal　prognostic　values.　Results　We　validated　the　heterogeneity　of　cells　in　the　tumor　microenvironment　and　revealed　the　presence　of　malignant　epithelial　cells　with　high　invasive　potential　within　primary　colonic　carcinomas.　Next,　the　＂stem-to-invasion　path＂　was　identified　through　constructing　a　branching　trajectory　with　cancer　cells　arranged　in　order.　A　multilayer　regulatory　network　considered　as　the　vital　factor　involved　in　acquiring　invasion　characteristics　underlying　the　path　was　built　to　elucidate　the　interactions　between　tumor　cell　and　tumor-associated　microenvironment.　Then　we　further　identified　a　novel　combinatorial　biomarker　that　can　be　used　to　assess　the　prognosis　for　CRC　patients,　and　validated　its　predictive　robustness　on　the　independent　dataset.　Conclusion　Our　work　provides　new　insights　into　the　acquisition　of　invasive　potential　in　primary　tumor　cells,　as　well　as　potential　therapeutic　targets　for　CRC　invasiveness,　which　may　be　useful　for　the　cancer　research　and　clinical　treatment.