Additionally, CDLC is also compared with the latest approach CEPPK (Li et al., 2014) which predicts essential proteins based on a fraction of known essential proteins. CEPPK provides us with a new angle to effectively discover essential proteins and the results show that CEPPK is an effective method. CEPPK selects a fraction of known essential proteins as a “seed” to predict the essentiality of other proteins. The number of “seed” essential proteins has only little influence on the prediction precision of CEPPK. Therefore, we only randomly select 50 and 100 known essential proteins for 20 times as samples to compare CDLC with CEPPK, respectively. Here, prediction precision means the proportion of true essential proteins out of a certain number of top ranked proteins (Li et al., 2014). To achieve a fair comparison, these “seed” essential proteins are excluded from the top ranked protein list before calculating the prediction precision of CEPPK and CDLC. As shown in Fig. 5, when 50 known essential proteins are selected to be seeds, the prediction precision of CDLC is higher than that of CEPPK no matter how many top ranked proteins are considered. Similar results can be found when 100 known essential proteins are selected to be seeds. In addition, the advantage of CDLC becomes more apparent with the number of known essential proteins increasing. So we can conclude that CDLC is more effective than CEPPK