#!/bin/bash
set -eux

CONVERTED_DIR=alt/v4/01_converted_data
GLOBAL_DIST_DIR=alt/v4/02_global_dist
TRAINDATA_DIR=alt/v4/03_traindata
TRAINED_PREDICTORS_DIR=alt/v4/04_trained_predictors
APPLY_DATA_DIR=alt/v4/05_apply_data

mkdir -p ${CONVERTED_DIR}
mkdir -p ${GLOBAL_DIST_DIR}
mkdir -p ${TRAINDATA_DIR}
mkdir -p ${TRAINED_PREDICTORS_DIR}
mkdir -p ${APPLY_DATA_DIR}

# Convert the DATA first...
chromimpute Convert -m H3K4me3 data marks.csv hg38.chrom.sizes ${CONVERTED_DIR} 
chromimpute Convert -m H3K36me3 data marks.csv hg38.chrom.sizes ${CONVERTED_DIR}
chromimpute Convert -m H3K9me3 data marks.csv hg38.chrom.sizes ${CONVERTED_DIR} 
chromimpute Convert -m H3K4me1 data marks.csv hg38.chrom.sizes ${CONVERTED_DIR} 
chromimpute Convert -m H3K27me3 data marks.csv hg38.chrom.sizes ${CONVERTED_DIR}
chromimpute Convert -m H3K27ac data marks.csv hg38.chrom.sizes ${CONVERTED_DIR} 


# Set up ComputeGlobalDist
chromimpute ComputeGlobalDist -m H3K4me3 ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR} 
chromimpute ComputeGlobalDist -m H3K36me3 ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR} 
chromimpute ComputeGlobalDist -m H3K9me3 ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR}
chromimpute ComputeGlobalDist -m H3K4me1 ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR} 
chromimpute ComputeGlobalDist -m H3K27me3 ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR} 
chromimpute ComputeGlobalDist -m H3K27ac ${CONVERTED_DIR} marks.csv hg38.chrom.sizes ${GLOBAL_DIST_DIR} 


# GenerateTrainData
chromimpute GenerateTrainData -dnamethyl wgbs_data/wgbs_data_files.tsv wgbs_data wgbs_data/header.tsv \
  -d 100 ${CONVERTED_DIR} ${GLOBAL_DIST_DIR} marks.csv hg38.chrom.sizes ${TRAINDATA_DIR} dna_methyl

# Train
samples=(D1 D3 D5 D7)
# samples=(D1 D2 D3 D4 D5 D6 D7 D8)
for sample in "${samples[@]}"; do
  chromimpute Train -dnamethyl wgbs_data/header.tsv ${TRAINDATA_DIR} marks.csv ${TRAINED_PREDICTORS_DIR} ${sample} dna_methyl
done

samples=(D1 D2 D3 D4 D5 D6 D7 D8)
for sample in "${samples[@]}"; do
  chromimpute Apply -dnamethyl wgbs_data/wgbs_data_files.tsv wgbs_data wgbs_data/header.tsv \
    ${CONVERTED_DIR} ${GLOBAL_DIST_DIR} ${TRAINED_PREDICTORS_DIR} marks.csv hg38.chrom.sizes ${APPLY_DATA_DIR} ${sample} dna_methyl
done

# IllegalArgumentException: No previously trained classifiers for mark dna_methyl were found available to load!
#         at ernst.ChromImpute.ChromImpute.executeApply(ChromImpute.java:2285)
#         at ernst.ChromImpute.ChromImpute.<init>(ChromImpute.java:1145)
#         at ernst.ChromImpute.ChromImpute.main(ChromImpute.java:8107)
# java.lang.IllegalArgumentException: No previously trained cla


# Note - I had to change L39 above to L40 above in order to get this to work.
# (i.e. using the same other files)...