How exactly Matlab performs the Lasso classification?

I am doing some research on Lasso classification method. I have a 40x15 dataset and I want to develop a binomial equation without dividing data into train and test set (because of small sample size).

I need some specific details on how exactly Matlab performs the LASSO (lassoglm). I know the general concept of how it works regarding removing insignificant terms by applying a penalty term to the coefficients, I also know about the CV and how it takes a subset of data and perform the analysis on that subset, but I still have some specific question about the process of LASSO in Matlab.

I want to know:

• in the case of not using CV, and just doing the B=lassoglm(X,Y), what are the values of B? Is it the matrix of X coefficients that I can use for the binomial equation?
• can I still be able to extract Index1SE and IndexMinDeviance to have the most significant terms contributing to my model without using CV?
• as CV finds the optimal value for lambda, when I, for example, use CV=10 how does the Matlab perform the lasso (lambda=100)? Does Matlab perform different analysis for each value of lambda? If so, HOW? and HOW it plots the Lambda-Deviance graph, having different error bars?
• and finally, can I just ignore CV and do simple Lassoglm classification, but still find the best, significant terms out of my 15 terms and then form a binomial formulation (having Intercept,C, and X coefficients)