Train a model from several samples

[guangingmai]

Hello, I have 10 samples from ocean, and I have co-assembled samples and assembled samples individully. Now, I want to train a model for co-assembled samples (all_contig.fa) and each assembled sample (S1_contig.fa ... S10_contig.fa) using the following code:
SemiBin2 train_semi \ -i S1.fa S2.fa ... S10.fa \ --data S1/data.csv S2/data.csv ... S10/data.csv \ --data-split S1/data_split.csv S2/data_split.csv ... S10/data_split.csv \ -c S1/cannot.txt s2/cannot.txt ... S10/cannot.txt \ --train-from-many \ -o train_output

1. Does this model can be used for all_contig.fa, S1_contig.fa ... S10_contig.fa?
2. Semibin2 with self-supervised option without the '-i' parameter. I have no idea how to modify the above code for self-supervised.
3. Does the result from the following codes same as multi-sample binning in S1 sample file?
   SemiBin2 single_easy_bin \ -i S1_contig.fa \ -b S1.sorted.bam ... S10.sorted.bam \ -o output
4. Does Semibin2 (version 2.1.0) set the self-supervised as default?