add fix for no specific label

[Jon-bioinfo]

if _select_one_most_specific has no prediction and appends '' in

CellO/cello/cello.py

Lines 738 to 743 in 7d37581

	for exp in binary_results_df.index:
	if exp in exp_to_select_pred_label:
	da.append(exp_to_select_pred_label[exp])
	else:
	# There was no prediction for this experiment
	da.append('')

that seems to break

CellO/cello/cello_predict.py

Lines 191 to 194 in 7d37581

	ms_results_df['most_specific_cell_type'] = [
	ou.cell_ontology().id_to_term[x].name
	for x in ms_results_df['most_specific_cell_type']
	]

so I add this check and give the label unknown.

[mbernste]

Thank you very much for addressing this issue! My only hesitation with this fix is that theoretically there should be a prediction for every sample... I'm not sure what is breaking that would cause a given cell type to be labelled with an empty string. While this fix may be a quick patch, it does not fundamentally solve the issue. Do you happen to have a dataset on which this error occurs and I can dig into the underlying problem?