To create the phenotype prediction model with PhenotypeSeeker, open the command-line and type in:
phenotypeseeker modeling data.pheno
Where "data.pheno" is an input text file containing tab separated lists of (1) sampleID's, (2) sample FastA/FastQ file addresses and (3) sample phenotype values (one or more column). The first line of the input file is header specifying the names of the columns. The names of the phenotype columns are used in the names of the output files of the corresponding phenotype analyses.
Example of "data.pheno" inputfile with binary resistance phenotypes for 5 antibiotics:
SampleID Address Ciprofloxacin Imipenem Meropenem Tobramycin Colistin
PA2_D4 /storage8/erkia/data/pseudomonas_genomes/PA2_D4.fasta 0 NA 0 NA 0
PA2_F7 /storage8/erkia/pseudomonas_genomes/PA2_F7.fasta 1 1 1 0 1
1D4 /storage8/erkia/pseudomonas_genomes/1D4.fasta 1 1 1 NA 1
6G2 /storage8/erkia/pseudomonas_genomes/6G2.fasta 1 1 1 1 1
PA4_A1 /storage8/erkia/pseudomonas_genomes/PA4_A1.fasta 0 0 0 0 0
6F4 /storage8/erkia/pseudomonas_genomes/6F4.fasta 0 NA 0 NA 0
5H2 /storage8/erkia/pseudomonas_genomes/5H2.fasta 1 1 1 NA 1
PA2_C6 /storage8/erkia/pseudomonas_genomes/PA2_C6.fasta 1 0 1 1 0
PA3_A4 /storage8/erkia/pseudomonas_genomes/PA3_A4.fasta 0 0 0 0 0
PA3_A10 /storage8/erkia/pseudomonas_genomes/PA3_A10.fasta 0 1 0 NA 0
Example of "data.pheno" inputfile with continuous resistance phenotypes for 5 antibiotics:
SampleID Address Ciprofloxacin Imipenem Meropenem Tobramycin Colistin
PA3_E8 /storage8/erkia/data/pseudomonas_genomes/PA3_E8.fasta 0.1250 2.0000 3.0000 4.0000 2.0000
7D2 /storage8/erkia/pseudomonas_genomes/7D2.fasta 2.0000 1.5000 3.0000 12.0000 0.5000
PA3_H11 /storage8/erkia/pseudomonas_genomes/PA3_H11.fasta 0.0940 4.0000 4.0000 8.0000 1.5000
PA2_H12 /storage8/erkia/pseudomonas_genomes/PA2_H12.fasta 0.0940 32.0000 3.0000 0.5000 2.0000
5E7 /storage8/erkia/pseudomonas_genomes/5E7.fasta 0.1250 3.0000 3.0000 1.0000 256.0000
PA3_H4 /storage8/erkia/pseudomonas_genomes/PA3_H4.fasta 0.1250 6.0000 0.0940 1.0000 1.0000
PA2_C4 /storage8/erkia/pseudomonas_genomes/PA2_C4.fasta 32.0000 32.0000 0.1250 0.2500 2.0000
PA2_A3 /storage8/erkia/pseudomonas_genomes/PA2_A3.fasta 16.0000 2.0000 0.2500 0.2500 NA
PA3_A9 /storage8/erkia/pseudomonas_genomes/PA3_A9.fasta 32.0000 4.0000 6.0000 0.5000 NA
6F7 /storage8/erkia/pseudomonas_genomes/6F7.fasta 32.0000 8.0000 32.0000 2.0000 1.0000
To limit the analysis on a selection of phenotypes present in "data.pheno", specify the phenotype columns with "--mpheno" option. For example, to analyse only the phenotypes of ciprofloxacin and meropenem from example "data.pheno" file, type in:
phenotypeseeker modeling data.pheno --mpheno 1 3
After launcing the PhenotypeSeeker modeling it starts counting the k-mers from input samples.
To save the temporary k-mer lists, the temporary repository "K-mer_lists" is created.
The process of k-mer counting results in the "k-mer_matrix.txt" file, where the number of each counted k-mer (rows) in each sample (columns) is recorded.
Then, if "-w" (weights) option is chosen it estimates the distances of samples with program mash, which output is saved in "mash_distances.mat".
Those distances are used to calculate GSC (Gerstein, Sonnhammer and Chothia) weights, which are considered in forthcoming analysis to correct for clonal population structure.
Then the "chi-squared_test_results.txt" or "t-test_results.txt" file is created depending on which type of phenotype (binary or continuous accordingly) were used.
The columns in "chi-squared_test_results.txt" represent (1) the tested k-mer sequences, (2) the chi-squared statistic values, (3) the pvalues of chi-squared statistics, (4) the numbers of samples with the specific k-mers and (5) the names of the samples with the specific k-mers.
Example of "chi-squared_test_results.txt" file:
ACCTCTGGGTGGCGAA 0.18 9.13E-01 3 | PA2_F7 PA2_G7 PA3_F7
AGCGCTACTCTTGATC 0.07 9.67E-01 8 | PA4_H4 PA3_G6 PA3_G4 PA3_A5 PA4_D2 PA3_G3 PA3_G9 PA3_G8
AAACTTTCGACACAGC 0.18 9.13E-01 3 | PA2_A2 PA2_A1 PA3_D6
GACCTGTCCCTCTCCA 0.18 9.13E-01 3 | PA2_B10 PA2_H6 PA2_C10
GCTCAATCGCTAAAGA 2.42 2.98E-01 2 | PA3_F10 PA3_C1
AAGTCGCTGGATTTCG 4.73 9.39E-02 7 | 1C9 PA3_A8 6E7 PA3_F7 PA2_F12 PA3_C7 PA3_A2
GGCACCCCGTTGGCCA 3.96 1.38E-01 9 | 6E5 1P4 PA3_B8 PA3_C9 PA2_B8 5I6 PA3_C4 PA2_F9 PA2_F4
AGCGCTGCTGCGCGAT 0.02 9.92E-01 2 | PA4_B3 PA4_H2
The columns in "t-test_results.txt" represent (1) the tested k-mer sequences, (2) the Welch's t-test statistic values, (3) the pvalues of t-test statistics, (4) the mean phenotype values of samples missing the specific k-mers, (5) the mean phenotype values of samples having the specific k-mers, (6) the numbers of samples with the specific k-mers and (7) the names of the samples with the specific k-mers.
Example of "t-test_results.txt" file:
CGAGCTCCCGGCA 24.86 3.14E-62 5.0 -0.54 4 | PA3_F5 PA2_F2 PA2_E2 PA2_G1
ATTGTCAAGGCGC -10.89 4.06E-11 -3.27 -0.38 3 | PA4_F4 PA4_B2 PA2_H6
TTGCGAAGATAAA 24.47 1.71E-61 5.0 -0.48 2 | PA2_G6 5I3
AGCAAGGTGAGTC -11.48 1.20E-23 -3.0 -0.39 3 | PA2_C8 PA3_C8 PA3_E1
CCATTTTTGTTAC -9.02 5.81E-10 -2.85 -0.38 4 | PA2_C2 PA2_A2 PA3_B5 PA2_A1
CGAACTATAACCC 24.47 1.69E-61 5.0 -0.48 2 | PA3_G8 PA3_G7
GTCGCCCTCTGGA -8.2 3.86E-06 -3.03 -0.37 4 | PA4_F4 PA3_H10 PA4_F1 PA3_H11
AATGCCTACGAAG -11.46 1.30E-23 -3.0 -0.4 2 | PA4_A4 PA4_A5
CTGTTGGATCCCA 6.88 2.79E-06 3.24 -0.68 13 | PA3_E8 PA2_D10 PA3_D1 PA3_C10 PA3_C3 PA3_D7 PA3_H7 PA3_D3 PA3_F9 PA2_E9 PA3_C5 PA2_H3 PA4_G1
CACAACCCCAACA -11.48 1.20E-23 -3.0 -0.39 3 | PA2_C8 PA2_C9 PA2_C4
The following file "k-mers_filtered_by_pvalue.txt" contains the subset of k-mers having the p-value lower than the p-value cut-off specified by "--pvalue" (with -B or -FDR) option(s).
Follows the regression analysis with top "--n_kmers" lowest p-valued k-mers from "k-mers_filtered_by_pvalue.txt" file.
The regression model is outputted in "log_reg_model.pkl" or "lin_reg_model.pkl" file depending on which type of phenotype (binary or continuous accordingly) were used.
The "summary_of_(log\lin)_reg_analysis.txt" contains the information about conducted regression analysis. For example the regularisation parameter chosen, the actual vs predicted phenotypes of test set samples and the model-evaluation metrics.
The columns in the following file "k-mers_and_coefficients_in_(log/lin)_reg_model.txt" represent (1) the k-mers used in regression model as parameter, (2) the regression model coefficients of k-mers, (3) the numbers of samples with the specific k-mers and (4) the names of the samples with the specific k-mers.
Example of "k-mers_and_coefficients_in_(log/lin)_reg_model.txt" file:
K-mer coef._in_lin_reg_model No._of_samples_with_k-mer Samples_with_k-mer
CAAGGCACTTTCC -2.68E-01 143 | PA2_F7 1D4 6G2 PA4_A1 6F4 5H2 PA2_C6 PA3_A4 PA3_A10 1P3 PA4_F2 PA4_F4 PA4_B2 PA2_C3 PA2_B4 6F2 6B5 PA3_B10 PA3_H10 PA2_E4 PA2_G7 PA4_A4 7P3 PA2_C8 PA3_C8 PA3_E1 PA2_D9 PA2_C2 PA2_B2 PA2_A2 PA2_D1 PA4_D4 PA2_B1 PA3_D10 PA3_F1 PA2_G8 PA2_A4 6G5 PA2_H12 PA2_H1 PA2_C9 PA4_G2 PA2_F8 PA4_A5 PA4_E2 PA4_F1 PA2_G12 PA3_G2 PA3_F6 PA3_E5 PA3_E10 PA3_B6 PA2_F6 PA2_B10 PA2_A10 PA2_A3 6G1 PA2_H6 PA3_F4 PA4_E4 PA4_B3 PA4_H4 PA4_H1 PA3_H11 PA4_E1 PA2_E1 PA2_A9 1D3b 6E9 6F3 PA3_H2 PA3_G6 PA3_G4 5C6 PA2_A1 PA2_A7 PA2_B3 PA2_D3 PA4_D1 PA2_F1 PA4_C1 PA2_G4 PA3_B3 PA2_E6 PA2_F3 1P4 6F7 PA3_F3 PA2_E8 PA3_C2 PA3_A6 PA3_A5 PA2_H4 PA2_C7 PA4_D2 PA3_F10 PA3_B8 PA4_H2 5D7 PA3_B4 PA2_D6 PA3_E8 PA3_E7 PA3_H9 PA2_D7 PA3_C9 PA2_B8 5F7 PA3_G10 5D4 PA4_C2 PA3_H5 PA3_D6 PA2_E7 5H6 PA3_A8 PA3_A11 PA3_A7 6E7 PA3_F8 6B9 PA3_H4 PA3_H3 PA3_G1 PA3_E4 PA3_G8 6A1 PA3_B1 PA3_A1 7D2 7D1 1D3a 1D6 5E7 PA2_C10 PA3_D5 PA2_B7 PA3_F7 PA2_F12 PA3_C7 PA3_A2 PA2_F4 PA4_G4
CTGCCGGGAACGA 0.00E+00 24 | PA2_B2 PA3_D8 5D4 6F9 PA3_D1 PA3_C3 PA3_D7 PA3_H7 6E7 PA3_G9 6B1 PA3_D3 PA2_G6 PA3_G8 PA3_C1 6A1 PA3_D2 PA3_C5 PA2_H3 PA2_F9 PA3_F7 PA2_F12 PA3_C7 PA3_A2
AACGCTTTCGGAC 0.00E+00 36 | 5B8 PA2_G3 7P1 PA2_F3 PA2_A8 PA2_E3 PA3_D8 PA3_B2 5C3 PA3_A8 6B7 6F9 PA3_D1 PA3_C10 PA3_C3 PA3_D7 PA3_B7 PA3_H7 6E7 6B9 PA3_G9 6B1 PA3_D3 PA2_G6 6A1 PA3_F9 PA3_D2 PA2_E9 PA3_G7 PA3_C5 PA2_H3 PA3_D5 PA3_F7 PA2_F12 PA3_C7 PA3_A2
CTCGGGGCGGAAA 0.00E+00 19 | PA3_D1 PA3_C3 PA3_D7 PA3_H7 6E7 PA3_G9 6B1 PA3_D3 PA2_F2 PA2_E2 PA3_D2 PA3_C5 PA2_H3 PA2_F9 PA3_F7 PA2_F12 PA3_C7 PA3_A2 PA2_G1
GCCTGTAGGTCGC 0.00E+00 69 | 1D4 6F4 PA3_A4 1P3 PA4_F4 PA4_B2 PA2_E4 PA2_B9 7P3 PA3_D10 5B8 PA2_G3 PA3_F1 PA2_F8 PA4_E2 PA2_H6 PA4_E4 PA2_F1 PA2_G4 PA2_A8 PA2_E3 PA3_B4 PA3_D8 PA4_C3 PA4_G3 PA4_D3 PA4_E3 PA4_F3 PA3_B2 5I6 5C3 PA2_D10 PA3_A8 6B7 PA3_D1 PA3_C10 PA3_C3 PA3_A7 PA3_D7 PA3_B7 PA3_H7 PA3_H8 6E7 PA3_G9 6B1 PA3_D3 PA3_F5 PA2_F2 PA2_E2 PA3_H3 PA3_G1 PA3_F9 PA3_D2 PA2_E9 PA3_G7 PA3_C5 PA2_H3 PA3_B1 PA3_A1 7D2 7D1 5I3 PA3_D5 PA3_F7 PA2_F12 PA3_C7 PA3_A2 PA2_G1 PA4_G1
TCGCGGTCTACGA 0.00E+00 42 | PA2_C9 6E5 PA3_B4 5D4 5I6 PA2_D10 6B7 PA3_D1 PA3_C10 PA3_C4 PA3_A11 PA3_C3 PA3_A7 PA3_D7 PA3_H7 6E7 PA3_F8 6B9 PA3_G9 6B1 PA3_D3 PA3_F5 PA2_F2 PA2_E2 PA2_G6 PA3_H3 PA3_G1 PA3_G8 PA3_C1 6A1 PA3_F9 PA3_D2 PA2_E9 PA3_C5 PA2_H3 5I3 PA2_B7 PA3_F7 PA3_C7 PA3_A2 PA2_F4 PA2_G1
CCTTGACCGAACG 7.17E-01 35 | 1D4 PA2_B9 PA2_B2 PA4_A5 1C9 PA4_C3 PA4_G3 PA4_D3 PA4_E3 PA4_F3 5D4 5I6 PA3_D1 PA3_C3 PA3_D7 PA3_H7 PA3_G9 6B1 PA3_D3 PA3_H4 PA2_G6 PA3_G8 6A1 PA3_D2 PA3_C5 PA2_H3 PA3_B1 PA3_A1 7D2 7D1 5I3 PA2_F9 PA3_F7 PA3_C7 PA3_A2
"PhenotypeSeeker prediction" predicts the phenotypes of input samples using the model previously created with "PhenotypeSeeker modeling".
Therefore the "PhenotypeSeeker modeling" outputfiles "(lin/log)reg_model.pkl" and "k-mers_and_coefficients_in(log/lin)_reg_model.txt" are essential to run the "PhenotypeSeeker prediction".
To predict the phenotypes of samples under study, open the command-line and type in:
phenotypeseeker prediction inputfile1 inputfile2
Where:
"inputfile1" is a text file containing tab separated lists of (1) sampleID's and (2) sample FastA/FastQ file addresses;
"inputfile2" is a text file containing tab separated lists of (1) the names of the phenotypes to predict, (2) corresponding model ("(lin/log)reg_model.pkl") addresses and (3) corresponding k-mer list ("k-mers_and_coefficients_in(log/lin)_reg_model.txt") addresses.
Example of "inputfile1":
PA3_H5 /storage8/erkia/data/pseudomonas_genomes/PA3_H5.fasta
PA3_D10 /storage8/erkia/pseudomonas_genomes/PA3_D10.fasta
PA3_A8 /storage8/erkia/pseudomonas_genomes/PA3_A8.fasta
PA4_F4 /storage8/erkia/pseudomonas_genomes/PA4_F4.fasta
PA4_B2 /storage8/erkia/pseudomonas_genomes/PA4_B2.fasta
PA4_A2 /storage8/erkia/pseudomonas_genomes/PA4_A2.fasta
5B8 /storage8/erkia/pseudomonas_genomes/5B8.fasta
PA3_F9 /storage8/erkia/pseudomonas_genomes/PA3_F9.fasta
PA3_G9 /storage8/erkia/pseudomonas_genomes/PA3_G9.fasta
PA3_D2 /storage8/erkia/pseudomonas_genomes/PA3_D2.fasta
Example of "inputfile2":
Ciprofloxacin /storage8/erkia/PhenotypeSeeker/log_reg_model_Ciprofloxacin.pkl /storage8/erkia/PhenotypeSeeker/k-mers_and_coefficients_in_log_reg_model_Ciprofloxacin.txt
Imipenem /storage8/erkia/PhenotypeSeeker/log_reg_model_Imipenem.pkl /storage8/erkia/PhenotypeSeeker/k-mers_and_coefficients_in_log_reg_model_Imipenem.txt
After launcing the "PhenotypeSeeker prediction" it starts counting the k-mers from input samples, followed by the detection of presence or absence of the model specific k-mers in each sample.
To save the temporary k-mer lists, the temporary repository "K-mer_lists" is created.
The data of model specific k-mers presence or absence in each sample is used to predict the phenotypes for samples.
The results of predictions are saved into "predictions_(phenotype_name).txt".
Example of "predictions_(phenotype_name).txt":
Sample_ID predicted_phenotype probability_for_predicted_phenotype
PA3_D8 1 0.82
PA3_E7 0 0.59
PA3_H9 0 0.92
PA2_D7 0 0.95
PA3_C9 0 0.93
PA2_B8 0 0.89
PA4_C3 1 0.88
PA4_G3 1 0.91
5F7 0 0.85