1. Indivumed # Tumor pairs (T) & normal control (N) #DifferentiationTumor Stage Lymph Node (LN) and vessels (LV) Metastases stageResectionGenderAge A3429T1/N1W1LN0/LV0R0M76 R78T2/N2W1LN0/LV0R0M60 R311T3/N3W3LN0/LV0R0M62 A2983T4/N4P3LN0/LV0R0F79 A2992T5/N5P3LN0/LV0R0F80 A3036T6/N6P4LN0/LV0R0M66 A2602T7/N7M1LN0/LV0R0F73 A2613T8/N8P4LN1/LV0R0M80 A2648T9/N9M3LN0/LV0R0M66 A2672T10/N10M2LN0/LV0R0M50 A1778T11/N11M3LN0/LV1R0M73 A1887T12/N12M3LN0/LV0R0F75 A1888T13/N13M2LN0/LV0R0M57 A2652T14/N14P3LN1/LV1R2F66 A2673T15/N15M4LN2/LV1R2F68 Supplementary Table S1. Tumor sample and patient information

2. Supplementary Table S1. Tumor sample and patient information ( to be continued) Indivumed # Tumor pairs (T) & normal control (N) #DifferentiationTumor Stage Lymph Node (LN) and vessels (LV) Metastases stageResectionGenderAge KC71T16/N16W1LN0/LV0R0F66 A3800T17/N17W2LN0/LV0R0F81 A3869T18/N18W2LN0/LV0R0F80 A2566T19/N19W3LN0/LV0R0M76 B1253T20/N20W1LN0/LV1R0F43 A3666T21N21W3LN0/LV0R0M83 R293T22/N22W3LN0/LV0R0F75 R428T23/N23W3LN0/LV0R0M74 Z320T24/N24W3LN0/LV0R0F70 Z322T25/N25W3LN0/LV1R0F80 A1976T26/N26M3LN1/LV0R0M64 A1995T27/N27M3LN0/LV0R0M60 A2019T28/N28M3LN0/LV0R0M84 A2076T29/N29M3LN0/LV0R0M65 A2081T30/N30M3LN0/LV1R0F78 A2082T31/N31M2LN0/LV0R0F72 A2099T32/N32M3LN0/LV0R0F51 A2103T33/N33M2LN0/LV0R0F65 A2106T34/N34M3LN0/LV0R0M33 A2156T35/N35M3LN1/LV0R0M77 Table. S1. Patient information was provided by Indivumed. Tumor and normal controls were numbered in the order they were sequenced. Differential stages (grade) were defined as well differentiated (W), moderately differentiated (M), and poorly differentiation (P). Lymph node (LN) metastases were defined as the following: LN0= no LN had metastases, LN1=1 to 3 LNs had metastases, LN2=4 or more LN had metastases. Lymphatic vessel (LV) metastases were defined as LV1. Resection is defined as tumor residue left after surgery with R0=no residue tumor, R1=microscopic residual tumor, R2=marcoscopic residue tumor. Tumor tissues had 60% tumor content plus desmoplasia and normal tissues were normal colon mucosa and submucosa with 0% tumor content at histological level. Normal tissues were minimum 5 cm away from tumors.

3. Supplementary information (S1) Fidelity of RNA-Seq analysis and identification of house keeping genes Since the regulation of 14 % of 270 miRNAs reported in this study is inconsistent with current references in CRC (3), the fidelity is very critical for evaluation studies. This variability may be caused by a number of factors including: use of different detection methods such as Q-PCR, microarray and RNA-Seq; differences in the storage of tissue used for the studies (eg, paraffin-embedded tissue versus frozen tissue); and lack of blood perfusion following resection and preservation, which is known to cause marked changes in gene. Thus, surmounting these challenges is difficult. Accumulating evidences suggest that ncRNA clusters may have one core promoter region and transcriptional start site, which are shared by all ncRNA genes within that cluster and thus are expressed within a single ncRNA transcript (16). In that regard, the trends and consistency of ncRNA cluster expression may be used in lieu of housekeeping genes as indicators of the accuracy of a quantitation study. We, therefore, sought to verify the fidelity of RNA-Seq by examining expression patterns of known clustered sequences including the snord12 (Fig. S1a) and snord114 clusters (Fig. S1b). Expression correlations of both snord12 cluster and snord114 cluster were strong since their coefficients (r) were all above 0.60. The relative expression levels of these clusters had similar trends in all 35 paired samples indicating the accuracy of RNA-Seq for non-coding RNA. Using ncRNA clusters as internal controls have not yet been reported. Moreover, commonly used ncRNA normalizers were found unsuitable in some quantification assays (36, 37). We specifically searched invariable genes in our first 15 CRC pairs data bank. From 5,171 sRNAs (including house-keeping genes: U6 families) and 1,009 miRNAs, we identified let7a and snord25 among having smallest relative variation. For let7a, the largest variation ratio across 15 pairs was 3.9 and the largest variation within a pair of tumor and normal tissues was 1.5. For snord25, the largest variation across 15 pairs was 2.5 and the largest variation within a pair was 2.4 (Fig. S1c). Based on these studies, we propose that, as the best candidates for house-keeping genes, let7a be used in studies of paired tumor and normal samples and that snord25 be used across numerous samples. Since let7a is downregulated in tumor, it is a biased normalizer in Q-PCR study of human primary CRC (Fig. S1d). The expression of traditional house keeping genes such as GAPDH and actin B were also variable between tumors and their matched normal controls (Fig. S1e). Thus without ideal house-keeping genes, it is theoretically difficult to verify RNA-Seq data with Q-PCR in paired matched CRC samples. Distinct profiles for mitochondrial t-RNAs and small nucleolar RNAs in locally invasive and metastatic colorectal cancer.

4. Fidelity of RNA-Seq data proved by snord RNA clusters (2) (Supplementary Fig. S1b) (Supplementary Fig. S1a)

5. miRNA and sRNA candidates for normalizers Biggest ratio between 30 samples: T11/N3=2.5. Biggest ratio within pairs: T6/N6=2.4 Biggest ratio between 30 patients: N3/T8=3.9. Biggest ratio within pairs: N3/T3=1.5 Biggest ratio between 2 patients : T15/T8=6.3. Biggest ratio within pairs: T3/N3=3.8 Biggest ratio between 2 patients: T7/N8=17.6. Biggest ratio within pairs: T7/N7=5.6 mir93mir16 Supplementary Fig. S1c

6. T/N=0.63T/N=1.27T/N=0.80(3)(1)(2) Supplementary Fig. S1d

7. Fig. S1. RPKMs of snord12 cluster (a) and snord114 cluster (b) were plotted in 35 CRC pairs. Correlation coefficient (r) was calculated by excel 2003. Since all rs were larger than 0.70 for snord12 cluster and snord114 cluster, there were strong positive linear relationship among the expression levels of genes in snord12 cluster and snord114 cluster in 35 CRC pairs. (c) Identification of ncRNAs as “house keeping” genes with smallest variation among first 15 pairs. ncRNAs were plotted by their RPKMs. (d) An example of using Let7a as a house keeping gene. (1) let7a expression in T7 pair. N7 had higher level of let7a. (2) mir378a expression in T7 pair. N7 had higher level of mir38a. (3) mir378a expression in T7 pair after normalization with let7a. N7 had “lower” level of mir378a. Therefore, the use of let7a as a housekeeping gene would be unreliable. (e) The expression profile of mRNA “house keeping” genes, GAPDH and actin B in first 14 CRC pairs. Expression of GAPDH was higher in 13 out of 14 tumors and expression of actin B was higher in 7 out of 14 tumors. Supplementary Fig. S1e

8. (Supplementary Fig. S2a)

9. Supplementary Fig. S2c Supplementary Fig. S2b

10. Supplementary Fig. S2d Supplemental Fig. S2. (a) 3D PCA of 22 MT-tRNAs (RPKM) in 35 CRC pairs. Four LN/LV positive tumors (T11Lv, T14LN, T15LN, T26LN) (boxed) and 2 LN/LV negative tumors (T12, T13) were clearly separately from remaining 29 tumors and 35 normal controls. (b and c) Identification of most up/down regulated sRNAs in tumors. All 35 tumors and 35 controls were grouped into 2 groups (T and N). Average T/N ratios were obtained, and log2 ratios of LN/LV positive /negative ratio were sorted from high to low as shown. (c) Nine MT-tRNAs were among top expressed 20 sRNAs. (d) Nineteen snoRNAs were among the 20 most downregulated sRNAs. (d) 3D PCA of 32 snoRNAs RPKMs in 35 CRC pairs. Twenty-eight (boxed) out of thirty-five tumors were separated from 35 normal controls and remaining 7 tumors.

11. Supplementary Table. S2a. 9 upregulated MT-tRNAs p-value MT-TF (phenyalanine)7.69E-06 MT-TL1 (leucine 2)0.0052 MT-TC (cysteine)0.013 MT-TL2 (leucine 2)0.015 MT-TY (tyrosine)0.025 MT-TI (isoleucine)0.026 MT-TE (glutamic acid)0.036 MT-TH (histidine)0.048 MT-TS1 (serine 1)0.169 13 upregulated snoRNAshost genesp-value snord12BZFAS13.83E-12 snora84NOL87.03E-12 snord78GAS55.06E-10 snord12CZFAS11.68E-09 snord86?5.29E-08 snord19BGNL37.91E-07 snord77GAS53.75E-06 snord12ZAS17.32E-06 snora71DAP1G15.22E-05 CTD265120.6SHF0.0014 snord43?0.008 snord49BTRPV20.07 snord1A?0.08 19 downregulated snoRNAshost genesp-value snord59AATP5B2.42E-08 snord114-1PWS6.65E-08 snord125ADTB17.03E-07 snord71AP1G17.65E-07 AP000318.1?9.22E-07 snora54RP21.38E-06 snord107PWS2.07E-06 scarna18TMEM167a1.25E-05 snord113PWS1.69E-05 snord114-14PWS2.96E-05 VTRNA1ZMAT22.96E-05 snord14-23PWS3.11E-05 snord64PWS4.96E-05 snord114-25PWS0.00011 snord116-7PWS0.0082 snord116-29PWS0.0086 snord83.9?0.0099 snord93RL12P100.108 Supplementary Table. S2b. 13 upregulated snoRNAs Supplementary Table. S2c. 19 downregulated snoRNAs

12. 9 upregulated Mt-tRNAs WMPWMP WMPWMP Supplementary Fig. S3a

13. WMPWMP WMP WMP

14. WMP

15. 13 Upregulated snoRNAs WMP WMP WMPWMP Supplementary Fig. S3b

16. WMP WMP WMPWM P

17. W MP WMP WMPWMP

18. WMP

19. 19 Downregulated snoRNAs WMPWMP WMP WMP Supplementary Fig. S3c

20. WMP WMP WMP W MP

21. WMP WMP WMPWMP

22. WMPWMP WMP W MP

23. WMP WMP WMPWMP

24. 9 upregulated 9 Mt-tRNAs Supplementary Fig. S3d

27. 13 Upregulated snoRNAs Supplementary Fig. S3e

31. 19 downregulated snoRNAs Supplementary Fig. S3f

36. Supplementary Fig. S3. Nine regulated MT-tRNAs (a), 13 upregulated (b) and 19 downregulated snoRNAs (c) were plotted by their RPKMs in 35 CRC pairs. W, M and P boxes represent well, moderately and poorly differentiated tumors respectively. Red arrows indicate LN/LV positive tumors. Nine upregulated MT-tRNAs (d), 13 upregulated (e) and 19 downregulated snoRNAs (f) were plotted by their average RPKMs in 35 CRC pairs grouped by well (w), moderately (m) and poorly (p) differentiated tumors (t) and their normal controls (n).

37. Ranks of deregulated sRNAs with smallest p-values in CRC Fig. S4. (a) Top twenty sRNAs with smallest p-values (paired t-test) in all 35 CRCs. (b) All examined 14 tumors had higher expression of ZNFX1-AS1 (host gene of snord12 cluster). (c) Expression correlation between snord12B and ZNFX1-AS1 in 14 tumors. There was a moderate correlation (r=0.66) between snord12B and ZNFX1-AS1 at RPKM ratio (T/N). Supplementary Fig. S4a. Supplementary Fig. S4b.Supplementary Fig. S4c.

38. Correlation between MT-tRNAs and miRNAs Supplementary information. S2a

39. Correlation between snoRNAs and miRNAs Supplementary information. S2b

40. Correlation of snoRNAs between miRNAs (to be continued) Supplementary information. S2b (to be continued 1)

41. Correlation between MT-tRNAs and mRNAs Supplementary information. S2c

42. Correlation between snoRNAs and mRNAs Supplementary information. S2d

43. Correlation between snoRNAs and mRNAs Supplementary information. S2d (to be continued 1 )

44. Correlation between snoRNAs and mRNAs (to be continued 1) Supplementary information. S2d (to be continued 2)

45. Correlation of snoRNAs and mRNAs (to be continued 2) Supplementary information. S2d (to be continued 3)

46. Supplementary information. S2d (to be continued 4) Supplementary information S2. Correlation between MT-tRNAs and miRNAs. (b) Correlation between MT-tRNAs and mRNAs. (b) Correlation between snoRNAs and miRNAs. (d) Correlation between snoRNAs and mRNAs. (correlation strength (r) >0.6). Correlation of snoRNAs and mRNAs (to be continued 3)

47. T7T5T16T19T20T22T27T30T31T35 up CTD26S77S12CS1AS12CS19CTD26S77S78S19 5.54.51.73.21.27.54.59.12.17.9 S12S19BS12BS77S78S77S71DS19BS12BS71D 3.64.11.63.21.24.84.18.61.83.8 S12BS101S12S71DS19S71DS19S19.2 S78 33.61.63.21.24.83.67.91.83.7 down S116-7S107S114-26VTRNA1S113-7S114-17S107 S114-26 -14-2.5 -3.2-2.6-3-2.7-2.3-3.8-13 S107S64AP00 S113-9S114-1S64AP00S114-14S107 -14-2.5 -5-4.3-3-2.9-3.2-3.8-13 S64S116-7VTRNA1S114-26S113-8AP00S113-9S114-16S114-1S64 -50-3.2-3.4-5-4.3-3-3.1-5.2-4.7-13 Type I: snoRNAs only CRC tumor fingerprint based on T/N PRKM ratio (1) Supplementary Table S3a

48. T1T2T3T10T13T11T12T9T15T4T14T6T8 up Mt-TS1S49AS77CTD26Mt-TL1 S71DCTD26Mt-TL1CTD26Mt-TYMt-TQS19B 4.5141747.95.58.720 3.75.1 9.1 S98S19Mt-TL1S19BS12BS71DMt-TL1S15AMt-TE Mt-TS1S19BMt-TS1 4.513103.55.748.55.99.83.74.54.88.6 S77Mt-TL1Mt-TEMt-TFS19BMt-TS2CTD26Mt-TFMt-TYRNY4Mt-TFMt-TEMt-TL1 3109.42.85.42.76.25.79.12.73.84.37.9 down S114-26 S71VTRNA1S93S107AP00S116-7S116-29VTRNA1S116-7 S64 -3.3-3.8-2.1-2.2-3.8-2.3-4.7-2.4-5.2-3-6.6-2.5-25 S114-23S114-14SC18mi3607AP00SC18 S114-14S116-7S64S107S123S116-29 3.3-3.8-4.5-2.2-3.8-2.5-5.5-2.6-6.6-3.7-9-3.4-25 S114-14S114-1AP00S125SC18S64VTRNA1S49BS64AP00S65S64S116-7 -3.4-4.7-7.1-2.8-6.2-2.7-9-3.4-9-5-16-7.6-25 Type II: MT-tRNAs + snoRNAs CRC tumor fingerprint based on T/N PRKM ratio (2) Supplementary Table S3a (to be continued 1)

49. T17T18T21T23T24T25T26T28T29T32T33T34 up S19S71DS77CTD26 Mt-TL1 S71DS19.2Mt-TL1 3.14.85.26.736.6135.13175.5 CTD26 S78Mt-TYS19Mt-TE Mt-THMt-TL1S78Mt-TFMt-TY 1.3344.51.84.154.52.61033.6 Mt-TL1Mt-TS1Mt-TS2Mt-TEMt-TL1Mt-TF Mt-TEMt-TYMt-TFMT-TS2Mt-TE 1.69.443.55.72.543.829.42.63 down S114-12AP00S114-14VTRNA1S49BS116-29S123S114-3S114-14AP00S107S14-26 -1.6 -4-2-5-2.6-3.2-4.5-3.5-3.3 -2.5 S114-3S116-17S114-1S49BS116-29S49BS107S114-26S113-9mir3607AP00S114-1 -1.6 -5-2.6-5-3.7-3.2-5-3.7 -3.3-2.6 S123S28AP00 S107AP00 S114-23S114-14S64 AP00 -2.7-2-25-4-5-3.7-4-5 -3.8-3.7-3.8 CRC tumor fingerprint based on T/N PRKM ratio (3) Supplementary Table. S3a (to be continued 2) Type II: MT-tRNAs + snoRNAs

50. CRC tumor fingerprint based on T-N PRKM difference (1) Type I: snoRNAs only Supplementary Table. S3b

51. CRC tumor fingerprint based on T-N PRKM difference (2) Type II: MT-tRNAs + snordRNAs Supplementary Table. S3b (to be continued 1) Supplementary Table. S3. Table. S3. (a) Ratio tumor fingerprint. (b) Subtractive tumor fingerprint. Ratio (T/N) and subtractive differences (T-N)) between each tumor pair were calculated and sorted from high to low. Three top up/downregulated genes were selected from each tumor. Two subtypes of tumors were identified based on snoRNAs only (type I) and MT-tRNAs/snoRNAs (type II).

52. Fig. S5. snord1114 cluster (a) and MEG3 (b) were plotted by their RPKM in 14 CRC pairs. There had strong positive correlation between MEG3 and snord114 cluster (r> 0.8). Supplementary Fig S5a Supplementary Fig S5b