altered training init to work with SLURM and corrected typo (BLOSUM --> BLOSUM62) in hyperparameters, specifically encoding options

fixed nonsensical import of subpackage in middle of script to point to mhc_rank instead of mhcflurry

trying to figure out why cleave_radius is an unexpected keyword arg in vector_encode even though vector encode has an explicit argument cleave_radisu...

Fixed big in argparse for make_architectures.py script. Created file with architecture options (1296) and saved as hyperparameters.json to ap_models dir

debugged architecture / training & created script to make models fitting all combinations of varied hyperparameters for tuning

added feature to flank_encoding.py, enabling pssm scores to be added as first position in residue encoding vectors

extract cleavage site, converted pssm from pd.Series and pd.DataFrames to np.ndarrays, enabled pssm calculation.

Added trim feature to seq_score of PSSM.py; Adjusted flanking_encoding.py to account for added encoding methods and for cleavage site extraction; Fixed bug in amino_acid.py so that correct encoding could be achieved

Added alignment method to EncodableSequences class in encodable_sequences.py. Next commit will inlcude commenting

Added PSSM retrieval for peptides. Included padding functionality for peptides shorter than max length using PSSM of X at middle residues of max length k-mer. Also, fully commented code and added docstring for class

Finalized / adjusted docstrings for all the methods of PSSM class. Still need to write docstrings for class itself.

Added X PSSM scores to CSVs. Also, updated PSSM.py, specifically: added docstrings, commented code, introduced assertions to functions, and enabled pssm retreival for series of flank seqences from length 0 to 5

merge