Uniformly shaped harmonization of gene expression profiles is central for the simultaneous comparison of multiple gene expression datasets. It is expected to operate with the gene expression data obtained using various experimental methods and equipment, and to return harmonized profiles in a uniform shape. Such uniformly shaped expression profiles from different initial datasets can be further compared directly. However, current harmonization techniques have strong limitations that prevent their broad use for bioinformatic applications. They can either operate with only up to two datasets/platforms or return data in a dynamic format that will be different for every comparison under analysis. This also does not allow for adding new data to the previously harmonized dataset(s), which complicates the analysis and increases calculation costs. We propose here a new method termed Shambhala‐2 that can transform multi‐platform expression data into a universal format that is identical for all harmonizations made using this technique. Shambhala‐2 is based on sample‐by‐sample cubic conversion of the initial expression dataset into a preselected shape of the reference definitive dataset. Using 8390 samples of 12 healthy human tissue types and 4086 samples of colorectal, kidney, and lung cancer tissues, we verified Shambhala‐2's capacity in restoring tissue‐specific expression patterns for seven microarray and three RNA sequencing platforms. Shambhala‐2 performed well for all tested combinations of RNAseq and microarray profiles, and retained gene‐expression ranks, as evidenced by high correlations between different single‐ or aggregated gene expression metrics in pre‐ and post‐Shambhalized samples, including preserving cancer‐specific gene expression and pathway activation features. © 2022 Wiley Periodicals LLC. Basic Protocol: Shambhala‐2 harmonizer Alternate Protocol 1: Linear Shambhala/Shambhala‐1 Alternate Protocol 2: Alternative (flexible‐format and uniformly shaped) normalization methods Support Protocol 1: Watermelon multisection (WM) Support Protocol 2: Calculation of cancer‐to‐normal log‐fold‐change (LFC) and pathway activation level (PAL)

151 Background: Colorectal cancer (CRC) is the fourth most common cancer worldwide with relatively poor patient survival. Transcriptome assay could be used to personalize CRC treatment thus complementing standard mutation analysis. Methods: We performed retrospective hybrid experimental and meta-analysis of CRC patient gene expression data with available progression-free survival (PFS) information and/or targeted drug response status. In total we analyzed 243 gene expression profiles from four publicly available (TCGA and three datasets from Gene Expression Omnibus GSE19860, GSE19862, GSE104645), and one experimental (PRJNA663280) patient cohorts. Each gene expression profile was analyzed using bioinformatic second-opinion platform Oncobox to calculate balanced drug efficiency scores (BES) to build personalized ratings of potentially effective targeted drugs. Area under the ROC curve (AUC) metric and Cox regression analysis were used to assess Oncobox capacity to predict tumor response and PFS, respectively. Results: Patients from GSE19860 (n = 12), GSE19862 (n = 14), GSE104645 (n = 81) received bevacizumab as monotherapy or in combination with chemotherapy as the nearest line of treatment after biopsy collection. Oncobox correctly classified treatment responders vs non-responders with AUC 0.94, 0.90 and 0.84, respectively. BES value was strongly associated with PFS (HR = 0.53, CI 0.33-0.84, log-rank test p-value 0.0057) in the GSE104645 cohort. However, BES was ineffective for predicting response and PFS after second-line (after biopsy collection) treatment with cetuximab. BES also predicted treatment response with AUC 0.74 in the TCGA cohort (n = 17) treated with 4 different targeted drugs. Thirty clinical outcomes were collected for 14 patients from our experimental cohort PRJNA663280. Patients were treated with 10 different targeted drugs. BES was an effective biomarker that could predict treatment outcomes with AUC 0.74 for all lines of therapy and 0.94 for the first line therapy (after biopsy), and could predict PFS after first-line treatment (HR 0.14, CI 0.027-0.73, log-rank test p-value 0.0091). Conclusions: Our results suggest that RNA profiling in tumor samples may be helpful for personalizing prescriptions of targeted therapeutics in CRC. Using recent biopsies is essential to obtain robust estimates of targeted drugs efficacy.